research paper on power system protection

• Search by keyword
• Search by citation

Page 1 of 7

Discrete space vector modulation and optimized switching sequence model predictive control for three-level voltage source inverters

This paper proposes a discrete space vector modulation and optimized switching sequence model predictive controller for three-level neutral-point-clamped inverters in grid-connected applications. The proposed ...

• View Full Text

Adaptive overcurrent protection scheme coordination in presence of distributed generation using radial basis neural network

The operational performance of conventional overcurrent protection relay coordination connected to a distribution network is adversely affected by the penetration of distributed generators (DG) at different bu...

Bi-level stackelberg game-based distribution system expansion planning model considering long-term renewable energy contracts

With the deregulation of electricity market in distribution systems, renewable distributed generations (RDG) are being invested in by third-party social capital, such as distributed generations operators (DGOs...

Low-carbon economic multi-objective dispatch of integrated energy system considering the price fluctuation of natural gas and carbon emission accounting

Natural gas is the main energy source and carbon emission source of integrated energy systems (IES). In existing studies, the price of natural gas is generally fixed, and the impact of price fluctuation which ...

Integrated risk measurement and control for stochastic energy trading of a wind storage system in electricity markets

To facilitate wind energy use and avoid low returns, or even losses in extreme cases, this paper proposes an integrated risk measurement and control approach to jointly manage multiple statistical properties o...

An improved constraint-inference approach for causality exploration of power system transient stability

Transient stability is the key aspect of power system dynamic security assessment, and data-driven methods are becoming alternative measures of assessment. The current data-driven methods only construct correl...

Tracking-dispatch of a combined wind-storage system based on model predictive control and two-layer fuzzy control strategy

To maximize improving the tracking wind power output plan and the service life of energy storage systems (ESS), a control strategy is proposed for ESS to track wind power planning output based on model predict...

“Adaptive virtual synchronous generator control using optimized bang-bang for Islanded microgrid stability improvement”

In this paper, a virtual synchronous generator (VSG) controller is applied to a hybrid energy storage system (HESS) containing a battery energy storage system and supercapacitor storage system for maintaining ...

Loss distribution analysis and accurate calculation method for bulk-power MMC

Accurate evaluation of power losses in a modular multilevel converter (MMC) is very important for circuit component selection, cooling system design, and reliability analysis of power transmission systems. How...

A tri-level programming-based frequency regulation market equilibrium under cyber attacks

Owing to their flexibility and rapid response, grid-connected distributed energy resources (DERs) are wielding growing influence in frequency regulation markets (FRMs). Nevertheless, compared with conventional...

Degradation state analysis of the IGBT module based on apparent junction temperature

The multi-chip parallel insulated gate bipolar transistor (IGBT) is the core device in large-capacity power electronic equipment, but its operational reliability is of considerable concern to industry. The app...

Suppression strategy for the inrush current of a solid-state transformer caused by the reclosing process

The automatic reclosing strategy is an effective measure to improve the reliability of a distribution network. It can quickly clear instantaneous faults in the grid. The traditional transformer has proven to b...

MPC-based LFC for interconnected power systems with PVA and ESS under model uncertainty and communication delay

In this paper, a cloud-edge-end collaboration-based control architecture is established for frequency regulation in interconnected power systems (IPS). A model predictive control (MPC)-based load frequency con...

Relay protection mirror operation technology based on digital twin

When conducting relay protection research, research costs can be significantly reduced if protection principle development, protection parameter verification and debugging can be carried out without relying on...

Dynamic economic evaluation of hundred megawatt-scale electrochemical energy storage for auxiliary peak shaving

With the rapid development of wind power, the pressure on peak regulation of the power grid is increased. Electrochemical energy storage is used on a large scale because of its high efficiency and good peak sh...

Sliding mode controller design via delay-dependent \(H_{\infty }\) stabilization criterion for load frequency regulation

Dual degree branched type-2 fuzzy controller optimized with a hybrid algorithm for frequency regulation in a triple-area power system integrated with renewable sources.

The uncertainties associated with multi-area power systems comprising both thermal and distributed renewable generation (DRG) sources such as solar and wind necessitate the use of an efficient load frequency c...

A fault segment location method for distribution networks based on spiking neural P systems and Bayesian estimation

With the increasing scale of distribution networks and the mass access of distributed generation, traditional centralized fault location methods can no longer meet the performance requirements of speed and hig...

An improved prediction method of subsequent commutation failure of an LCC-HVDC considering sequential control response

Subsequent commutation failure (SCF) can be easily generated during the first commutation failure (CF) recovery process in a line-commutated converter-based high voltage direct-current system. SCF poses a sign...

Two-stage stochastic-robust model for the self-scheduling problem of an aggregator participating in energy and reserve markets

This paper addresses a two-stage stochastic-robust model for the day-ahead self-scheduling problem of an aggregator considering uncertainties. The aggregator, which integrates power and capacity of small-scale...

Fault location of untransposed double-circuit transmission lines based on an improved Karrenbauer matrix and the QPSO algorithm

Some double-circuit transmission lines are untransposed, which results in complex coupling relations between the parameters of the transmission lines. If the traditional modal transformation matrix is directly...

Data-driven next-generation smart grid towards sustainable energy evolution: techniques and technology review

Meteorological changes urge engineering communities to look for sustainable and clean energy technologies to keep the environment safe by reducing CO 2 emissions. The structure of these technologies relies on the ...

Typology and working mechanism of a hybrid power router based on power-frequency transformer electromagnetic coupling with converters

The power router (PR) is a promising piece of equipment for realizing multi-voltage level interconnection and flexible power control in the future distribution power grid. In this paper, a hybrid PR (HPR) topo...

State-of-health estimation of lithium-ion batteries based on electrochemical impedance spectroscopy: a review

Lithium-ion batteries (LIBs) are crucial for the large-scale utilization of clean energy. However, because of the complexity and real-time nature of internal reactions, the mechanism of capacity decline in LIB...

Optimal PV array reconfiguration under partial shading condition through dynamic leader based collective intelligence

This paper applies the innovative idea of DLCI to PV array reconfiguration under various PSCs to capture the maximum output power of a PV generation system. DLCI is a hybrid algorithm that integrates multiple ...

Age-of-information-aware PI controller for load frequency control

Open communication system in modern power systems brings concern about information staleness which may cause power system frequency instability. The information staleness is often characterized by communicatio...

Circuit breakers in HVDC systems: state-of-the-art review and future trends

High voltage direct current (HVDC) systems are efficient solutions for the integration of large-scale renewable energy sources with the main power grids. The rapid development of the HVDC grid has resulted in ...

A novel hybrid cybersecurity scheme against false data injection attacks in automated power systems

The conventional power systems are evolving as smart grids. In recent times cyberattacks on smart grids have been increasing. Among different attacks, False Data Injection (FDI) is considered as an emerging th...

Analysis of low voltage ride-through capability and optimal control strategy of doubly-fed wind farms under symmetrical fault

Given the “carbon neutralization and carbon peak” policy, enhancing the low voltage ride-through (LVRT) capability of wind farms has become a current demand to ensure the safe and stable operation of power sys...

Transient synchronous stability analysis and enhancement control strategy of a PLL-based VSC system during asymmetric grid faults

The stability of a voltage source converters (VSC) system based on phase-locked loop (PLL) is very important issue during asymmetric grid faults. This paper establishes a transient synchronous stability model ...

Hierarchical under frequency load shedding scheme for inter-connected power systems

Severe disturbances in a power network can cause the system frequency to exceed the safe operating range. As the last defensive line for system emergency control, under frequency load shedding (UFLS) is an imp...

Two-stage distributionally robust optimization-based coordinated scheduling of integrated energy system with electricity-hydrogen hybrid energy storage

A coordinated scheduling model based on two-stage distributionally robust optimization (TSDRO) is proposed for integrated energy systems (IESs) with electricity-hydrogen hybrid energy storage. The scheduling p...

Battery energy storage-based system damping controller for alleviating sub-synchronous oscillations in a DFIG-based wind power plant

This paper presents the issue of the Sub-synchronous resonance (SSR) phenomenon in a series compensated DFIG-based wind power plant and its alleviation using a Battery Energy Storage-based Damping Controller (...

Distribution network state estimation based on attention-enhanced recurrent neural network pseudo-measurement modeling

Because there is insufficient measurement data when implementing state estimation in distribution networks, this paper proposes an attention-enhanced recurrent neural network (A-RNN)-based pseudo-measurement m...

Adaptive H ∞ event-triggered load frequency control in islanded microgirds with limited spinning reserve constraints

Using an islanded microgrid (MG) with large-scale integration of renewable energy is the most popular way of solving the reliable power supply problem for remote areas and critical electrical users. However, c...

Strategies for improving resilience of regional integrated energy systems in the prevention–resistance phase of integration

The construction of integrated energy systems can help improve energy efficiency and promote global energy transition. However, in recent years, the occurrence of extreme natural disasters has brought certain ...

Graph representation learning-based residential electricity behavior identification and energy management

It is important to achieve an efficient home energy management system (HEMS) because of its role in promoting energy saving and emission reduction for end-users. Two critical issues in an efficient HEMS are id...

Single-ended protection method for hybrid HVDC transmission line based on transient voltage characteristic frequency band

Hybrid high-voltage direct current (HVDC) transmission has the characteristic of long transmission distance, complex corridor environment, and rapid fault evolution of direct current (DC) lines. As high fault ...

Sensing as the key to the safety and sustainability of new energy storage devices

New energy storage devices such as batteries and supercapacitors are widely used in various fields because of their irreplaceable excellent characteristics. Because there are relatively few monitoring paramete...

Jointly improving energy efficiency and smoothing power oscillations of integrated offshore wind and photovoltaic power: a deep reinforcement learning approach

This paper proposes a novel deep reinforcement learning (DRL) control strategy for an integrated offshore wind and photovoltaic (PV) power system for improving power generation efficiency while simultaneously ...

Fault identification scheme for protection and adaptive reclosing in a hybrid multi-terminal HVDC system

A fault identification scheme for protection and adaptive reclosing is proposed for a hybrid multi-terminal HVDC system to increase the reliability of fault isolation and reclosing. By analyzing the "zero pass...

Comparative framework for AC-microgrid protection schemes: challenges, solutions, real applications, and future trends

With the rapid development of electrical power systems in recent years, microgrids (MGs) have become increasingly prevalent. MGs improve network efficiency and reduce operating costs and emissions because of t...

Time–frequency multiresolution of fault-generated transient signals in transmission lines using a morphological filter

The ongoing transformation of electrical power systems highlights the weaknesses of the protection schemes of traditional devices because they are designed and configured according to traditional characteristi...

Publisher Correction: Reactive power optimization of a distribution network with high-penetration of wind and solar renewable energy and electric vehicles

The original article was published in Protection and Control of Modern Power Systems 2022 7 :51

Voltage imbalance mitigation in an active distribution network using decentralized current control

Voltage imbalance (VI) is caused by the difference in connected single-phase load or generation in a low voltage distribution network (DN).VI increase in a smart distribution grid is due to the current practic...

Three-stage day-ahead scheduling strategy for regional thermostatically controlled load aggregators

Thermostatically controlled loads (TCLs) are regarded as having potential to participate in power grid regulation. This paper proposes a scheduling strategy with three-stage optimization for regional aggregato...

A multi-energy inertia-based power support strategy with gas network constraints

An integrated energy system with multiple types of energy can support power shortages caused by the uncertainty of renewable energy. With full consideration of gas network constraints, this paper proposes a mu...

Review of sub-synchronous interaction in wind integrated power systems: classification, challenges, and mitigation techniques

Emerging sub-synchronous interactions (SSI) in wind-integrated power systems have added intense attention after numerous incidents in the US and China due to the involvement of series compensated transmission ...

Detection of false data injection attacks on power systems using graph edge-conditioned convolutional networks

State estimation plays a vital role in the stable operation of modern power systems, but it is vulnerable to cyber attacks. False data injection attacks (FDIA), one of the most common cyber attacks, can tamper...

Solar-PV inverter for the overall stability of power systems with intelligent MPPT control of DC-link capacitor voltage

This paper demonstrates the controlling abilities of a large PV-farm as a Solar-PV inverter for mitigating the chaotic electrical, electromechanical, and torsional oscillations including Subsynchronous resonan...

• Search Menu
• Sign in through your institution
• Advanced Articles
• Editor's Choice
• Author Guidelines
• Publish with us
• Submission Site
• Open Access
• Self-Archiving Policy
• About Clean Energy
• About the National Institute of Clean and Low-Carbon Energy
• Editorial Board
• Instructions for Reviewers
• Advertising & Corporate Services
• Journals Career Network
• Journals on Oxford Academic
• Books on Oxford Academic

Article Contents

Introduction, 1 methods and challenges addressed, 2 ai techniques, 3 application of ai in the power sector, 4 conclusions, 5 future scope, authors’ contributions, conflict of interest statement, data availability.

• < Previous

Applications of artificial intelligence in power system operation, control and planning: a review

• Article contents
• Figures & tables
• Supplementary Data

Utkarsh Pandey, Anshumaan Pathak, Adesh Kumar, Surajit Mondal, Applications of artificial intelligence in power system operation, control and planning: a review, Clean Energy , Volume 7, Issue 6, December 2023, Pages 1199–1218, https://doi.org/10.1093/ce/zkad061

• Permissions Icon Permissions

As different artificial intelligence (AI) techniques continue to evolve, power systems are undergoing significant technological changes with the primary goal of reducing computational time, decreasing utility and consumer costs and ensuring the reliable operation of an electrical power system. AI techniques compute large amounts of data at a faster speed than numerical optimization methods with higher processing speeds. With these features, AI techniques can further automate and increase the performance of power systems. This paper presents a comprehensive overview of diverse AI techniques that can be applied in power system operation, control and planning, aiming to facilitate their various applications. We explained how AI can be used to resolve system frequency changes, maintain the voltage profile to minimize transmission losses, reduce the fault rate and minimize reactive current in distributed systems to increase the power factor and improve the voltage profile.

The demand for advanced research and technology has steadily increased in the sector of electric grids [ 1 ]. Automation and intelligent technology have become widely used in response to development demands over time. Traditional research methods are quickly becoming insufficient to enable data scientists and researchers to keep up with any global challenges that artificial intelligence (AI) may be able to assist us in solving and uncovering important insights among the billions of pieces of data scattered throughout power systems. AI can handle large amounts of data and utilize them to make power system operations, control and planning more efficient. The use of AI technology in power systems has been investigated and debated in related areas and has resulted in more study material and certain outcomes, which are reviewed in this paper. The demand for advanced research and technology has constantly risen in the sector of electricity grids. The application of AI technology to the automation of power system control can improve the efficiency of electrical automation management, mitigate the risk of accidents and ensure smooth operation of the power system over an extended period [ 2 ]. Evaluating the use of AI technology in power systems requires a comprehensive analysis of existing research in the field of artificial intelligence and its related industries.

The power system is a network consisting of three components: generation, distribution and transmission. In the power system, energy sources (such as coal, sunlight, wind, nuclear reactions and diesel) are transformed into electrical energy [ 3 ]. There are different power systems, such as solar power systems, wind power systems, thermal power plants [ 4 ], nuclear power plants, geothermal power plants, etc. All power systems have different structures and equipment for the generation of electricity [ 5 ]. The basic structure of a power system includes:

(i) generating substation;

(ii) transmission substation;

(iii) sub-transmission substation;

(iv) distribution substation.

In power system problem-solving, conventional approaches such as practical numerical optimization methods (e.g. lambda iteration and Newton–Rapson methods) have been used. Optimization problems are non-linear and, with the various constraints included, these optimization problems become slow and complex. So, several AI techniques are discussed here to solve many optimization issues with less computation time. Furthermore, experiments were conducted to identify which backpropagation algorithm would give the most efficient and reliable network training [ 6 ]. The systematic approach was introduced [ 7 ] for developing a recurrent neural network (RNN) that could effectively predict the dynamic behaviour of a pilot-scale entrained-flow gasifier. The RNN was trained using a data set consisting of input and output data collected from a dynamic read-only memory (ROM) system established in a previous study. To evaluate the accuracy of the RNN, comparisons were made with computational fluid dynamics models and experimental data obtained from the pilot-scale gasifier. The findings demonstrated that the RNN surpassed the performance of the dynamic ROM model, showing strong predictive capabilities to capture the transient behaviour of the gasifier. Additionally, this was achieved while significantly reducing the computational resources required. Power systems are one of the main study topics for the advanced development of AI. The use of AI algorithms in power plants has been under continual investigation since the advent of the expert system technique [ 8 ]. However, in most situations, issues such as long cycle times, complex computation and difficulty in learning arise with classic AI methods. In recent years, efficiency has increased significantly with the continued advancement of AI algorithms. The multi-source model of heterogeneous large data has gradually developed, the data volume continues to grow, and new possibilities and problems are being created with the use of AI in power systems. AI encompasses several technologies such as expert systems, pattern recognition, genetic algorithms (GAs) and neural networks. By incorporating AI into the automation of power system control, it has the potential to enhance the efficiency of electrical automation management, mitigate the risk of accidents and ensure long-term smooth operation of the power system. Machine learning (ML) has also found extensive applications in predicting the properties of rechargeable battery materials, particularly electrolyte and electrode materials, as well as the development of novel materials, according to relevant research. The breadth of ML applications will grow steadily as ML technology advances and new unique issues emerge in the research of rechargeable battery material [ 9 ]. Although ML has shown considerable potential in modelling complex systems, its implementation introduces new challenges. These challenges include difficulties associated with accessing relevant and reliable data sets, and addressing inaccuracies in model predictions needs to be addressed before ML can be widely deployed. To effectively implement ML on a large scale, it is crucial to have compelling evidence of its effectiveness in diverse areas such as manufacturing processes, energy generation, storage and management. Furthermore, the availability of commercial software and a skilled workforce specialized in the relevant domains is essential [ 10 ]. ML has proven to be beneficial in creating data-driven models that accurately correlate material properties with catalytic performance, including activity, selectivity and stability. As a result, there have been advancements in the development of effective design and screening criteria for solid-state catalysts with desired properties [ 11 ]. Nevertheless, there are still difficulties in applying existing ML algorithms to accurately predict catalyst performance or devise strategies for designing high-performance catalysts. The review emphasizes recent advancements in ML applied to solid heterogeneous catalysis, as well as the limitations and constraints faced by ML in this field, and also discusses some of the prospects for using ML effectively in the design of solid heterogeneous catalysts. Successful uses of ML in short-term hydrothermal scheduling will strengthen the link between real operations and issue formulation, and prepare the hydropower sector for autonomy by identifying the need for and availability of autonomous systems now and in the future. In this research, a review of the state of the art of ML applications for the hydroelectric sector was offered [ 12 ]. Digital technologies have a significant impact on energy market services and the safety of residents and energy consumers, particularly in smart homes. Sustainable smart home networks can improve energy efficiency, utilize local renewable energy, decarbonize heating and cooling systems, and promote responsible electric vehicle charging [ 13 ]. The next decade is crucial for achieving ambitious global CO 2 reduction targets and the decarbonization of buildings is a major challenge. Water-efficient development and resilient homes are essential for coping with impacts of climate change. Research on sustainability and energy efficiency is vital to improve the quality of life in the face of climate change [ 14 ]. Several important aspects were highlighted [ 15 ] with respect to the current state and prospects of smart homes. It was recognized that despite the increasing prevalence of smart homes and the growing familiarity with them, there were still significant obstacles that researchers must address to achieve widespread adoption [ 16 ].

One of the technical hurdles highlighted is the diversity of manufacturers and devices, each with varying charging systems, frequencies and communication methods [ 17 ]. This fragmentation can hinder interoperability and compatibility between devices and systems. The greater acceptance of smart home technologies poses a significant challenge. This review highlights the crucial task of convincing consumers about the safety and reliability of these technologies. An approach [ 18 ] was presented to develop prediction models that were capable of identifying faults and malfunctions in power equipment, demonstrating their effectiveness in predicting the progression of degradation phenomena. The challenges were discussed [ 19 ] associated with predicting the technical condition index of the equipment and determining the probability of its current state having defects. This research contributes significantly to the advancement of predictive analytics tools in the industry, enabling proactive maintenance of equipment. ML and data-driven approaches exhibit significant promise in the field of predictive analysis within power systems, especially in the context of smart grids. These methods can efficiently analyse the vast amounts of data collected from smart meters and other devices in real time, facilitating optimized energy flow in an increasingly renewable-energy-focused landscape [ 20 ]. They offer advantages such as improved accuracy, cost reduction and improved efficiency. However, certain challenges must be overcome, such as ensuring the availability of high-quality data and managing the potential risk of information overload [ 21 ].

The articles selected for review are based on different parameters and selection criteria. The shortlist is based on parameters such as duration, analysis, comparison and applications, as listed in Table 1 . The challenges investigated for power system operations, control and planning in the article are as follows and a diagram visualizing the domains of the power sector along with the AI techniques used and their application is presented as Fig. 1 . Power system operation [ 22 ] includes the total power requirement that must reliably meet the real-time generation, including transmission losses. The problems involved in this task are economic load dispatch (ELD), power flow, unit commitment and generator maintenance schedule.

Selection criteria for shortlisted research papers

Visual depiction of power sector domains, their application and AI techniques used

The complex and large design of the power system is presented [ 23 ] and interferences in the power system are a problem. When a large interference occurs, control tasks are needed to find the disturbed area, control the impact caused and bring the process to normal form. Heuristic solutions are non-linear and hence are not designed to deal with fast-occurring disturbances. Therefore, many control optimization techniques such as voltage control (VC), power system stability control and load frequency control are discussed to address this problem.

Power system planning has an arrangement of a power system that is complex and large with many parts such as flexible alternating current transmission system (FACTS) devices and distribution systems. The major goal of least-cost planning is to optimize the components required to deliver enough power at a minimal cost. Many factors such as FACTS placement and demand are given importance in the expansion of power system planning. Reactive power optimization, distribution system planning (DSP) [ 24 ] and capacitor placement are the optimization problems considered in this task [ 25 ].

2.1 Artificial neural network

In AI, a set of inputs is transformed into an output using a network of neurons. A neuron produces a single power by simply operating its input in the same way as a processor [ 26 ]. The working group of neurons and the pattern of their connections may be utilized to build computers with real-world issues in model recognition and pattern categorization. As the human brain processes are replicated, input signals are processed using mathematical operations utilizing artificial neurons.

The network consists of neurons organized in layers and connected to ensure information input–output flows [ 27 ]. By using what is known as the activation function, in layer ‘i’, each neuron is linked to the ‘i+1’ layer of all neurons. The input signals for a specific neuron originate from all neurons in the prior layer and their excitation power changes to govern the degree of signal reaching each neuron [ 28 ].

In several scientific disciplines, such as medical diagnosis, voice, pattern recognition, etc., artificial neural networks (ANNs) are utilized. The ANN is a computing system partly based on biological neural networks, expressed by linked nodes (artificial neurons), correctly structured in layers that are found in human or animal brains. All artificial neurons are linked and are able, employing their connections (synapses) to send signals, generally real values, which result in an output computed according to the original input, depending on the sizes allocated to all neurons [ 29 ].

ANNs are recognized as data-mining approaches capable of modelling several independent characteristics with dependent functions in non-linear functions. ANNs may predict a future value of a dependent variable after training with a comparable sample, replicating the learning process of a human brain [ 30 ].

In turn, a difference in signal strength affects the activation of the neuron and, as a result, signals that are transmitted to other neurons as shown in Fig. 2 .

Simplified diagram of the artificial neural network

Input layer—distribute other units but does not process the data.

Hidden layer—the ability to map the non-linear problems is provided through hidden layers.

Output later—the output units encode the value to be assigned to this instance.

2.2 Adaptive neuro-fuzzy interference system

The adaptive neuro-fuzzy interference system (ANFIS) creates an input/output data set whose membership function parameters are modified with the minimum square method type or the backpropagation algorithm by itself, using a fuzzy interference system [ 31 ]. This modification has helped the fuzzy system learn from the data it models. By applying hybrid learning, ANFIS utilizes a systematic approach to determine the optimal distribution of membership functions, enabling effective mapping of the relationship between input and output data [ 32 ]. The ANFIS architecture combines ANNs with fuzzy logic, making the modelling process more structured and less dependent on expert knowledge. This inference system is constructed using five layers in its basic form. Each ANFIS layer has several nodes defined in the layer specification using the node function. The current layer inputs from the preceding layer nodes are collected. The structure of the ANFIS is shown in Fig. 3 [ 33 ].

Simplified diagram of an artificial neuro-fuzzy interference system

The fuzzy-inference method involves organizing empirical information in a professional manner, which presents challenges in quantifying it through membership functions (MFs) and fuzzy rule bases [ 34 ]. Additionally, neural networks possess learning capabilities. From top to bottom, they are very adaptable in their system set-up and have great parallel processing and fault tolerance. The theories for neural network neuro-fuzzy systems are actively explored in several areas [ 35 ].

The utilization of a neuro-fuzzy system, which emulates human learning and decision-making abilities, can lead to varying model performances compared with traditional mathematical approaches. The process of rule generation and grouping in a neuro-fuzzy inference system, adapted to the specific model, can be approached through a grid-based methodology, known as the ANFIS.

2.3 Fuzzy logic

To identify the fuzzy set from which the value comes and the degree of membership within that set, fuzzy logic systems base their choices on input in terms of variables generated from the member functions [ 36 ]. The variables are then combined with IF–THEN language requirements (fuzzy logic rules) and a fluid implication is used to answer each rule [ 37 ]. The response of each rule is weighted according to the confidence or degree of the inputs of each rule and the central part of answers is computed to provide a suitable output and achieve the compositional rule of deference. Now, the design of fuzzy logic systems is not a systematic approach. The easiest way is to subjectively define member functions and rules with a human-operated system or an existing controller and then test the design for the right output. If the design fails the testing, the MFs and/or rules should be changed. Recent investigation directions involve the creation of fuzzy logic systems that can learn from experience.

Currently, only published findings can create and modify fuzzy control rules based on experience [ 38 ]. Among them can be Scharf’s self-organizing robotic control system [ 39 ] by using a performance matrix to change the rule matrix and alter the rules that constitute the management strategy. Another intriguing example is the Sugeno fuzzy vehicle, which can be trained to turn and park itself. Instead of a membership function, the effect of a rule is viewed as a linear equation of the process state variables. Through optimization of least-squares performance indices using a weighted linear regression system, the challenge is simplified to a parameter estimate. Although these approaches provide promising outcomes, they are subjective, somewhat heuristic and depend on trial and error for the choice of MFs. Thus, the ability to learn neural networks can offer a more promising approach to fuzzy logic systems [ 40 ]. As shown in Fig. 4 , the fuzzy logic system consists of four parts: knowledge base, fuzzification, inference and defuzzification. On the basis of the fuzzy constants provided, the process converts given inputs to the fuzzification stage. Based on the knowledge base, the inference is made. Then, in the defuzzification stage, every fuzzy output is mapped to complex output MFs [ 41 , 42 ].

Fuzzy logic system

2.4 Ant colony optimization

Ants exhibit a behaviour in which they remember and follow a specific path between their colony and a food source. They achieve this by leaving pheromone trails during their food search [ 43 ]. When other ants come across these pheromone trails, they start to follow them. Because the increased presence of the chemical on the path has the effect of attracting more ants to follow it, the ants will emphasize the pheromone trail. To find the best solution to the problem studied, ant colony optimization (ACO) builds multiple iterative solutions. The objectives of [ 44 ] were to evaluate the features of the search area for problems and to use this knowledge to address the solution process. The solution–construction process is a sequential decision-making process due to parametric stochastic decisions. An ACO algorithm depends on a sequence of learning of the parameters used in decision-making to reach a global policy that provides optimum solutions for a particular situation [ 45 ]. The parameters of the learning object are considered pheromones and are called variables of the pheromones.

An ACO method contains a stochastic local search technique to organize the routing pathways that artificial ants can determine. These ants co-operate together through indirect information exchange to create the best and shortest route. The concept of the ACO is taken from the food search characteristic of the true colony in an intelligent optimization algorithm and how the ants work together in this difficult job. It can be expected that the ACO finds the quickest route from nest to food according to the biological study of the ants. The ant pheromone distribution technique is termed staggered, in which information is shared with other ants indirectly. Pheromone updates are the basis of the ACO algorithm. These pheromone updates depend on the pheromone and the number of ants that work best. Natural ants can determine the quickest route based on their best knowledge and a strong pheromone trace. The shortest path is inversely proportional to the amount of pheromone and length of the path using an ACO method. The following is a step-by-step explanation of the algorithm replicating these properties [ 46 ]. The pseudocode for ACO is shown in Table 2 .

Pseudocode of the ant colony optimization algorithm

Set pheromone pathways: The algorithm starts by setting the initial pheromone pathways in the search space of the problem. These pathways act as a guide for ants to navigate and find solutions.

Generate a random ant population: Next, the algorithm generates a population of random solutions (ants) to start searching for the optimal solution.

Choose the optimal position: Each ant then uses a combination of pheromone information and heuristics to determine the next step (position) to take. The objective is to find the position that maximizes the target function.

Get the finest search ant: After all ants have completed their search, the algorithm selects the best ant, i.e. the one with the highest value of the target function.

Restore the trail of pheromone: The pheromone trail of the best ant is then updated to reinforce its path, encouraging other ants to follow it.

Check the end condition: The algorithm repeats the above steps until a stopping criterion is met.

End: The algorithm concludes when it satisfies the stopping condition and provides the best solution discovered.

2.5 Artificial bee colony optimization

The artificial bee colony (ABC) optimization imitates bee behaviour. A colony of bees is made up of onlookers, scouts and worker bees [ 47 ]. Artificial bees are flown in this system in a multidimensional search room and, depending on the experience they have gained and based on their next partner experience, the used bees pick their food sources and bees to change positions. Scout bees pick their food sources at random without any experience. Each food source avoids the probable solution to the problem under discussion [ 48 ]. The number of bees employed is as high as the food sources, each being a site, which is currently being used or as many solutions as individuals [ 49 ]. This procedure is continued until the ABC optimization meets a stop criterion.

ABC_Optimization (n, m, k)

population <- initialize (n, m, k)

global_best <- assign_random_food_source(population[m])

while! stop_criteria_met ()

for bee in population

fitness <- calculate_fitness (bee. food_source)

if fitness > global_best. fitness

global_best <- bee. food_source

for the bee in population

bee. update_food_source (global_best, bee. next_partner)

update_food_source (global_best, next_partner)

prob <- random_probability ()

if prob < experience

food_source <- global_best

else if prob < experience + next_partner. experience

food_source <- next_partner. food_source

food_source <- random_food_source ()

Initialization phase:

Initialize the x i j solution population in the j domain parameter. The exact description may be used for that purpose:

where x m a x j is the upper bound of the parameter j and x m i n j is the lower bound of the parameter j.

Worker bee phase:

Each worker bee uses a formula to identify and assess a food source v i j representative of a location such as a food source in her x i j memory. Each worker offers information about their food source to onlookers who select a food source website based on information collected from their bees while they wait at the hive according to Equation (2) :

If x k is a randomly picked solution, j is a parameter randomly selected and β i j is a random integer within the [–a, a] range. A greedy selection between v i and x i is applied after the production of a new solution v i ⁠ .

Onlooker bee phase:

There is a reference previously to the proportion of the amount of a food source to its location in the solution. Onlookers are positioned at food sources using a selection strategy based on fitness, such as the way of selecting the roulettes wheel. New solutions x i based on pi are picked to assess the new solutions v i and new solutions v i for spectators are created. The hired bees between v i and x i receive a greedy selection.

Scout bee phase:

Former workers who lost their resources start scouting randomly for food supplies. Every colony of bees has scout bees. The scouts have little instruction when looking for food. They mostly focus on finding food. Artificial bees can find the available answers rapidly. ABC decides that the artificial scout is the bee whose food supply has been lost or whose profitability has fallen below a specific level of profitability. The control parameter that decides the class is the withdrawal criterion or the ‘limit’. After a predefined number of attempts, a worker bee leaves an unimproved solution that is a source of food. The number of tests necessary to release the answer is determined by ‘limits’.

2.6 Particle swarm optimization

Particle swarm optimization (PSO) is a population-based evolutionary computational technique that is employed to address stochastic troubleshooting. It belongs to the category of swarm intelligence and is founded on social and psychological principles. PSO provides valuable insights into engineering applications and contributes to their development [ 50 ]. Social impact and social learning make cognitive consistency possible for the person. People may resolve issues by talking to people and by changing their ideas, attitudes and behaviour; they can usually be portrayed as people moving in a socio-cognitive space towards one another. But PSO has certain inconveniences such as global convergence; unlike some other optimization algorithms, PSO does not have a guarantee of global convergence, which means that it may not find the true optimal solution. To address this drawback, a novel PSO and a chaotic PSO are designed to tackle energy-system optimization issues efficiently. The analysis of the problem of unit commitment within the regulated system leads to the examination of UCP (uniform customs and practice for documentary credits) inside the deregulated market. The overall profit, execution time and convergence criteria are compared between various approaches.

One element is the current velocity of the particle v ( t ) ⁠ . Another is the optimum position Y ∗ ( t ) to approach the particle. The third factor is that the community or sub-community is optimally informed by Y ∗ ∗ ( t ) [ 51 ]. In each iteration step, the particle speed is modified to Y ∗ ( t ) and Y ∗ ∗ ( t ) ⁠ . Meanwhile, the random weight is independently allocated to the V i ⁠ , Y ∗ ( t ) and Y ∗ ∗ ( t ) ⁠ . The speed and position are updated following Equations (3) and (4):

In the given equation, v k +1( i , j ) represents the velocity of the particle in the i -th particle and j -th dimension at iteration k  + 1.

The weight factor ω determines the extent to which the previous velocity influences the new velocity.

v k ( i , j ) denotes the velocity of the particle in the i -th particle and the j -th dimension at iteration k .

C 1 and C 2 are the learning parameters that determine the influence of the personal best and global best solutions, respectively.

r and 1 and r and 2 are randomly generated numbers within the range of [0,1].

P bes t k ( i , j ) represents the personal best position of the i -th particle in the j -th dimension achieved thus far.

Y k ( I , j ) represents the current position of the i -th particle in the j -th dimension.

G b es t k signifies the global best position discovered by all particles up to the present iteration.

The flow chart for PSO is shown in Fig. 5 .

Flow chart of particle swarm optimization

2.7 Regression model

The research model [ 52 , 53 ] can be defined using Equation (5) :

where Y represents the dependent variable; this refers to the indication of respondent i ’s willingness to adopt smart home technology and their level of flexibility in terms of demand for technology j. β refers to the intercept. X 1 ij ,..., Xnij are dichotomous predictors included in the model. εij represents the random error term.

Building on Equation (5) , the level 2 model can be formulated as follows:

In Equations (5) and (6) , u 0 j ,..., u 1 j represent the random effects. W 1 j and W 2 j correspond to grand-mean centred and uncentred variables, respectively.

These equations are utilized in research to describe the relationships between the dependent variable, predictors and random effects. Equation (5) serves as the core model equation, capturing the influence of the predictors on the dependent variable while accounting for random error. Equations (5) and (6) extend the model by specifying the relationships and random effects associated with the intercept and predictor coefficients at the level 2 analysis. Collectively, these equations offer a comprehensive framework to analyse the variables that impact the acceptance of smart home technology and the adaptability of demand within the specific research context.

2.8 Regression and classification problems using AI

The RNN is a variation of the neural network frequently employed in the power systems domain to address regression and classification problems that involve sequential data. Unlike direct neural network models, the structure and operating principle of the RNN differ significantly [ 54 ]. In an RNN, the input data are fed to the model sequentially at each time step ( t ), as shown in the signal propagation diagram. At each step, the current state ( output ) is calculated by considering the current input data and the previously computed state. This iterative process continues for a fixed number of steps ( n ) until the desired output (predicted value) is achieved or until all input data ( input ) have been processed [ 55 ].

The propagation of signals in the RNN model is illustrated by the values assigned to each hidden state (hidden). These hidden states are calculated using the previous hidden state ( hidden  − 1) and current input data ( input ) [ 56 ]. hiddent = (〈 w hidden , hiddent – 1〉 + 〈 w input , input 〉) Here, σ () represents the activation function (such as the sigmoid function, hyperbolic tangent or rectified linear unit (ReLU)), while w hidden and w input are the weights for the hidden and input states, respectively.

The output value at each calculation step output is obtained by taking the dot product of the weights associated with the output state and the values of the hidden state, similar to a regression equation: output = 〈w output , hiddent 〉 [ 57 ].

During training, the initial stage involves calculating the output signal, after which the error function is calculated to determine the discrepancy. For regression problems, it is common to utilize the square root of the standard deviation between the output of the RNN and the values from the response space ( y t ):

Applying the chain rule, the gradient of the error functional is calculated. The weight coefficients ( w ij ) are adjusted in a manner that reduces the functional, following the direction of decreasing values, until it reaches the minimum value or the training iterations reach the predetermined limit. It is important to note that the weights associated with the hidden state of the RNN ( w hidden ) remain unchanged after propagating the error from each output ( output ). Conversely, the coefficients w output and w input change at each step of the gradient [ 58 ].

3.1 Operation of the power system

ELD is the process of assigning the generation output to the generation unit to supply the system load fully and economically. The whole engaged generating unit produces total electricity costs to minimize energy. The main concern of the ELD problem is to reduce the overall fuel cost. This is achieved by generating electricity in a way that optimizes the use of resources and reduces the overall cost of power production in the electricity system. Multiple generators provide enough total output to meet the consumer requirements in a typical power system. The production costs of each generating unit in the electricity system are different, as the producing units are not the same distance from the loading unit. Over the years, several AI approaches have been created to address this challenge. After simulation, authors concluded that using the genetic algorithm technique to solve the ELD problem can result in a lower overall cost of electricity production, but may also result in higher emissions. However, choosing a solution with a higher cost may result in minimum emissions. Transmission losses are usually neglected when they are small. However, for long-distance transmission in large and interconnected networks, transmission losses become significant and have an impact on the optimal distribution of power generation. It is possible to operate the same multitasking system with a better voltage profile and with evolutionary calculation technology, the cost and emission value of the best compromise.

The proposed results of the simulation of the ANN emphasize that the results are indiscernible from conventional methods, although the time used by neural networks is less than conventional methods. The number of generators increases the prediction error because there is a lot of input and output data to be learned. To monitor the performance, neural networks have been modelled. The authors have performed the simulations with many generation units having ramp rate limits and prohibited operating zones as constraints and the resultant performance is compared with ANN, GA and ACO techniques, but the ABC technique gives better outputs with fast convergence. The greedy selection procedure and the timely abandonment of the used food sources contained in ABC give it this potential. The basic operations of ABC optimization prevent solutions from stopping and make the algorithm more exploitative.

3.1.2 Generator maintenance scheduling

Generator maintenance scheduling (GMS) is a complicated combinatorial optimization issue for a power provider. Mathematical approaches include traditional ways to tackle the GMS issue. To evaluate the needed objective function, a mathematical model approach employs a trial-and-error procedure. Mathematical approaches even fail to provide viable answers as in some circumstances the operator needs to rely on certain assumptions and models that may not accurately reflect real-world conditions. In some cases, operators may need to be involved to provide additional input and expertise to make informed decisions about maintenance schedules. In addition, there may be unpredictable factors, such as equipment failures or changes in demand, that cannot be accurately accounted for by mathematical models alone. Maintenance is a preventive outage program for generating units within a certain time horizon in a power system. In the event of a range of various specification generating units in the energy system and several limitations to produce a sustainable and practical solution, maintenance planning becomes a difficult challenge. The maintenance planning of the generators is done for time horizons of different lengths. Short-term maintenance plans from 1 hour to 1 day are crucial to the daily operations, engagement and operational planning of power plants. Medium-term planning is necessary for resource management between 1 day and 1 year. Long-term planning from 1 to 2 years is crucial for future planning. An examination is being conducted to resolve some AI methods, including simulated developments, neural networks and GAs. The application of the genetic algorithm through case research shows that suitable GA parameters are safeguarded, as well as issue coding and development functions. The use of integer encoding decreases the velocity of the genetic search method investigation. By using integer encoding, the algorithm needs to perform additional operations to convert solutions into integers, which can slow down the search process. Planning the generation of power remains a barrier to competent solution technology and a difficult optimization problem. The challenge in power generation planning lies in finding the optimal balance between cost and efficiency, while also considering factors such as environmental impact, reliability and security of supply. The answer to the difficulties in generation planning consists of finding the UC (unit commitment) at every point in the programming period for each generator in one power system. An electrical system must be defined in each planning interval for each power generator for the decisions and levels of output. The solution process must be addressed concurrently for binary decisions and continuous variables. Generation difficulties with scheduling are typically quite narrow and combined. Match swarm optimization approaches have been used to achieve viable schedules within a specified time. The study found that an optimization-based approach using PSO provided better results than a GA or an evolutionary strategy. Data from the actual power system were used to evaluate the performance of the different optimization techniques.

UC is properly scheduled for the ON/OFF status and the genuine generator power outputs of the system. To satisfy a high number of system limitations and decrease the overall fuel cost at every time interval, a spinning reserve is necessary (spinning reserve refers to the additional generation capacity that is available and running but not actively supplying power to the grid). UC meets the expected load requirements in advance. To implement UCs, medium-term load forecasting using ANN was used. The neural network structure was trained through learning and parameter learning. Total operational expenses under 24 hours were used for the assessment criterion. The study demonstrates the effectiveness of the proposed approach by comparing the performance of the ANN-based load forecasting model with traditional methods such as linear regression and time-series forecasting. The results show that the ANN-based load forecasting model significantly improves the accuracy of load forecasting and reduces the scheduling cost by reducing the number of units needed for scheduling. The study also highlights the importance of considering the uncertainty and variability of load demand in UC scheduling and suggests that ANN-based load forecasting models can be a useful tool for achieving more efficient and reliable scheduling in power systems. Locational marginal prices have been evaluated through a trained ANN. The findings show that the current technique gives a different UC mechanism. To develop unit commitment, the PSO technique is used. On implementation, with the increasing size, the execution time is also increasing. To accelerate the PSO, a convergence repair method is also implemented.

3.1.4 Optimal power flow

Optimal power flow (OPF) is a highly important technique to identify the optimum control parameter settings that enhance or decrease the intended target function, but also under a variety of limitations. An essential instrument to design and operate a power system is the issue of optimum power flow to identify the best parameter settings that can maximize or minimize the intended goal function within specific limitations. Voltage and reactive controls, also called OPD, are an OPF sub-problem that seeks to reduce overall transmission loss by resuming the reactive power flow. Optimal reactive power dispatch is a non-linear solution for the issue of blending integers since some control variables such as tap ratios for transformers, shunt capacitor outputs and reactors are distinct.

The alternate strategy for mitigating the problem of GA-ANN is set out in this article. A collection of ANN networks is trained offline in specified system quantities to work on a general OPF issue. To choose the appropriate ANN inputs, the k-mean clustering technique is utilized. When learning the functions correctly, ANNs can easily estimate the associated results with great precision.

The ANFIS develops the input/output data set fuzzy-inference system (FIS) that matches the membership (adjustment) parameters with a backpropagation or minimum square process type. This update allows you to learn from the fuzzy systems data IEEE 39 bus system implementations and simulated software from the power world are utilized for the formation of the ANFIS. The results indicate that the ANFIS offers solutions as accurate as conventional ones. It takes less time, though, and it works quickly. Some additional papers on the application of AI in the operation of power systems are presented in Table 3 .

Applications of artificial intelligence in the operation of a power system

3.2 Control of the power system

The main objective of a voltage controller power system is to maintain the voltage profile within a defined limit, thus minimizing transmission losses and avoiding cases of voltage instability [ 81 ]. The VC system consists of three levels of hierarchical control: AVR (automatic voltage regulator), tertiary voltage control (TerVC) and secondary voltage control (SecVC). AVR is aimed at controlling the voltage of buses that are equipped with reactive power sources (e.g. synchronous, sync, static var compensators and STATCOM (static synchronous compensator)). Actions are carried out locally at this control level. SecVC is used to monitor the voltage on a specific bus that controls a cargo bus.

In situations in which there is hardware present in the vicinity that modifies the reference point for the AVR, the control level typically operates at a slower pace compared with the AVR control level. SecVC is responsible for identifying VC regions and their correlation with individual load buses. To accommodate varying power system conditions, SecVC must demonstrate flexibility in adjusting the control regions to accommodate all grid conditions. On the other hand, TerVC determines the optimal reference value for voltage grids at each load bus. The objective is to minimize power loss, optimize reactive power and maintain a minimum load release or reservation. TerVC is usually updated every 30 minutes to 1 hour.

The backward error propagation algorithm trains the multilayered feedforward perception. The minimum singular value method analyses the static voltage collapse. The procedure uses a minimum voltage stability evaluation time once the network training is complete. For monitoring voltage collapse, complementary methodologies of neural networks and expert systems would be combined for use in the application [ 82 ].

GA is an iterative optimization technique with several solutions from the candidates (known as a population). In the case in which there is no knowledge of the problem field, then the GA starts to look for solutions from a random population. The appropriate coding (or display) must first be defined to solve the problem. A fitness function should also be defined so that every coding solution is given a figure of merit. If parents are not satisfied with the termination condition, for reproduction, they must be picked [ 83 ]. They are then joined to generate offspring through reproduction and, to refresh the population of candidate solutions, crossover and mutation operators are utilized. Typically, in a basic genetic algorithm, three operators are involved: selection, crossover and mutation. These operations are performed to generate new offspring, individuals and subsequent generations. The same process is repeated with the new generation until the desired criteria are met. The approach of this method is used to teach swarming at the beginning of PSO. In this case, 10 control variables are used for the ANN input. The neuron and its prejudice are 11. A hidden layer consists of this group of neurons and biases. Ten outputs/goals are available. These objectives are achieved by using the optimal value of PSO. The last outputs are the initializations in the time-varying non-linear particle swarm optimization (TVNL-PSO). The steps are as follows:

PSO is used to take the ANN input; the weight value is applied at random;

the ANN input and partition in a cached layer are weighed and then activated by the sigmoid binary function;

weighting of the output in the hidden layer and activation of the linear function;

to optimize the reactive power and VC by TVNL-PSO, the ANN output is transmitted as a starting initialization value.

3.2.2 Power system stability control

The stability of a power system is a feature that allows it to remain under a balance in normal operating conditions and retrieve an acceptable balance after a change. Margins of stability can be seen to decline throughout the world [ 84 ]. We highlight three of the many reasons for this:

The inhibition by economic and environmental constraints of further transmission or construction. Therefore, power systems must be operated with lower safety margins.

Restructuring of the electricity industry. The restructuring process reduces the margins of stability, as power systems do not co-operate effectively [ 85 ].

Increased complexity of power systems multiplies the compulsive properties. These include large, non-linear oscillations; frequency differences between weakly binding energy-system areas; interactions with saturated devices.

Fuzzy logic endeavours to address problems by emulating human reasoning, allowing optimal decision-making based on available information. It can also be employed to regulate the stability of un-modelled systems. To achieve improved performance, a fuzzy logic (FL) controller is combined with a PID (proportional–integral–derivative) controller. In this particular scenario, the fuzzy logic control adjusts the gains of the PID controller based on the power system.

A fuzzy logic controller primarily consists of four major parts: fuzzification, fuzzy rule base, fuzzy inference and defuzzification. FACTS have proven to be extremely promising for increasing performance under stable conditions. The most promising FACTS device is a unified power flow controller (UPFC). Three control factors can be adjusted: bus voltage, reaction line and phase angle between two buses. The power flow should be redistributed across lines while a stable state is preserved. It can also be utilized to increase damping when low frequencies are damped temporarily.

Power system stability control.

Load frequency control as defined by the controllable generator power output control in a prescribed area resolves system frequency changes, two-line loadings or interactions to maintain an interchange with other regions within the fixed limit or scheduled system frequency [ 86 ]. The traditional proportional–integral (PI) controller is the most widely used among different types of load frequency controllers. The PI controller can be easily implemented and provides a faster response, but its performance decreases when unwanted disturbances, such as load change dynamics, increase the difficulty within the system. In this paper, less computing is required for the non-linear autoregressive-moving average-L2 (NARMA-L2) control architecture. Plant output, reference and control signals are included. The controller is therefore taught to monitor the output of the reference model. The model network that updates controller settings predicts the effect of the change in plant performance. Some additional papers on the application of AI in the control of power systems are presented in Table 4 .

Applications of AI in control of a power system

3.3 Planning of power system

DSP plays a crucial role in enhancing reliability and minimizing costs for both utilities and consumers. Electric power distribution networks are a fundamental component of the electrical power system. In general, transport networks are denser and more complex than those that provide transformer stations [ 110 ]. Automating previously manual jobs increases with distribution networks becoming more complex. New tools are known as advanced automation functions that support the operation of such networks. These functions enable the network operator to effectively address issues that arise. Furthermore, the reconfiguration of distribution networks is essential to identifying optimal solutions that align with the operator’s requirements and constraints, ensuring a secure and economically optimized electricity supply.

The optimal design of a distribution network is not a fixed solution, but rather a process that involves considering various technically feasible options and using improvement tools to make the best decision based on factors such as demand, reliability of power transmission and network structure. All potential paths are initially identified with uploaded system data and then the energy-loss cost calculation applies for each identified path forward/backward sweeping load flow technique. For the distribution of power, the minimum energy-loss path is chosen. The optimal selection of the branch conductor of the radial system is done using optimization of PSO. In this case, parameters such as power loss, voltage profile and capital investment depreciation improve optimization. These parameters are used as optimization criteria to determine the optimal branch conductor that minimizes power loss, improves the voltage profile and reduces capital investment depreciation. The PSO algorithm iteratively updates the position of each particle in the search space based on its own experience and the experiences of its neighbours. The algorithm continues until a global optimum is found or a stopping criterion is met. The final solution produced by the PSO algorithm represents the optimal branch conductor that meets the optimization criteria [ 111 ]. The optimization of the PSO results in the optimal conductor and the best substation, the positioning of the optimal conductor is selected and then the optimal substation power distribution is achieved.

A multi-target algorithm was proposed that uses a fluid optimization technique to handle contradictory targets [ 112 ]. The plan formulation and the algorithm include a multi-target function that uses battery energy storage systems (BESSs) and traditional resources to select the best planning option. The microgrid BESS has been receptive to power management and improvement in power quality. The proposed algorithm is based on the fuse-based decision-making processes of the Mamdani-type FIS and Bellman–Zadeh approach.

In this paper [ 113 ], two algorithms, namely the mixed-integer linear program (MILP) and GA, are compared for the design of a radial distribution system feeder. The main objective is to minimize total investment and operational outages while maximizing system reliability. The study aims to evaluate and compare these two optimization techniques in terms of their optimality, complexity and time requirements. A unique aspect of the optimization model is the consideration of operational costs associated with failures, which are directly linked to the design of the system. The fault rate and defect cost at each loading point are updated based on the proposed configurations. It is crucial to determine which method produces superior results in terms of optimality, complexity and time efficiency.

The GA technique is used to build the algorithm for optimizing distribution networks. The fundamental concept is the growth of the genetic operator population (selection, crossover and mutation). These are used to generate a fresh population from the previous generation throughout each generation procedure. In GA, a single chromosome shows each person. This chromosome corresponds, according to the graph theory, to a radial distribution network configuration or a particular graph twist. The chromosome group is the population. Randomly, an initial group is created as a first step in the implementation of GA. Then the encoding is applied to each chromosome. In this study, only closed branches represent the network topology. A true coding strategy was used to match each gene to the edge of the coagulation tree [ 114 ].

3.3.2 Reactive power optimization

As the demand of electricity increases and new lines are built, the environment and the unforeseen power fluid in the lens are reduced, it is generated by the current scenario. Effect reactive compensation control improves voltage, reduces energy loss and improves system performance under stable and dynamic conditions in weak nodes [ 115 ]. Because the complexity of power systems is constantly increasing and the network components are constantly being loaded, abnormal operating conditions such as voltage can occur more often. Therefore, it is obvious that the power system needs adequate reactive power and VC.

In pursuit of intelligent theory development, a combination of fuzzy logic and ANNs is used to determine the control strategy for transformer taps and capacitors. However, due to the increasing complexity of control variables, rapid optimization becomes challenging [ 116 ]. To address this issue, a genetic algorithm is utilized, which tackles problems associated with incorporating regulatory time as penalty terms in the objective function and determining appropriate penalty factors that affect algorithm performance. When regulatory time is a constraint, the optimization objective focuses on minimizing the total energy loss during the dispatch period. GAs, inspired by natural selection mechanics and genetics, such as inheritance, mutation and recombination, are utilized (also referred to as crossover) to optimize the solution.

The PSO method can be used for handling FACTS devices in power systems. Various reactive power problem objectives and different solutions are addressed in the interconnected power system. Solutions and comparative analyses using the FACTS device, differential evolution (DE) and PSO algorithms are presented under various loading conditions [ 117 ]. The algorithm proposed [ 118 ] employs DE to minimize generator fuel costs on FACTS devices. Additionally, the authors discuss the hybridization of DE and PSO (DEPSO) to overcome the maximum load limit. The control of reactive flow is addressed using fuzzy sets and a fuzzy feature optimization technique is introduced to optimize reactive power. The utilization of fuzzy linear programming offers an effective approach to calculating reactive power, aiming to minimize active power loss and maximize the voltage stability margin. The paper [ 119 ] explores the combination of fuzzy and GA approaches for FACTS shunt controller placement and sizing. Lastly, the focus of [ 120 ] is on the integration of fuzzy systems with GM algorithms and the PSO algorithm to tackle the OPF problem and optimize control variables. In this paper, the authors [ 121 ] focus on fluid-based reactive and voltage controls to reduce actual loss of power.

3.3.3 Capacitor placement

There are some advantages if capacitors are placed optimally, including various factors, such as maximizing energy and reducing peak power loss through the introduction of a condenser in an electrical distribution system. In the paper, a novel adaptive modified firefly algorithm is presented to address the optimal capacitor placement problem in power systems. This optimization problem involves identifying the best positions and sizes of capacitors in a power system, in order to enhance voltage stability, minimize energy losses and improve the power factor. The proposed algorithm combines the firefly algorithm with adaptive parameter settings and introduces a unique crossover operator to enhance both convergence speed and solution quality. The authors conduct evaluations on a test system and compare the performance of the algorithm with other optimization methods. The results demonstrate the effectiveness of the proposed approach in finding optimal solutions and highlight its superiority in terms of solution quality and computational efficiency compared with other algorithms [ 122 ]. Losses occur due to reactive currents in the distribution system and are therefore minimized in the right places. Shunt capacitors are used, depending on their use. A capacitor is used to improve the voltage profile, reduce losses and increase the power factor [ 123 ].

Elbaz et al. [ 124 ] have been using ANN techniques to control both capacitor banks and voltage regulators. The ANN has many input connections and all inputs are combined to determine the output capacity. The purpose of the capacitor search algorithm was to reduce total active losses in the distribution system by utilizing the capacitor banking search to address the capacitor placement problem. The operation of the ant colony was proposed to address problems related to the installation of the capacitor. The fuzzy method uses variables such as angle, current and voltage, etc. A degree for a set and fuzzy variable is determined by MFs. This degree changes from zero to one that takes zero or one as opposed to the classical methods [ 125 ]. A fuzzy logic-based algorithm is developed to minimize line loss for the placement of condensers in a radial system. The fuzzy expert system identifies the capacitor candidate nodes by compromising the possible reduction in loss between the condenser system and the voltage level. This paper [ 126 ] presents a fluffy approach to identifying the appropriate places for capacitor placement. In the design of a seamless logic to determine the optimal placement of capacitors, two main objectives are taken into account. These objectives include (i) minimizing actual power loss and (ii) maintaining voltage within acceptable limits. Fuzzy member functions are used to model voltage and power loss indices for nodes in the transmission system [ 127 ]. The suitability of capacitor placement for each node in the distribution system is then determined. This is achieved using a set of rules within the FIS. Nodes can be fitted with capacitors of greatest appropriateness.

In [ 128 ], a method based on GAs is used to identify the optimal locations and sizes of capacitors in the distribution network. The capacitor sizes are considered discrete and known variables placed on the buses of the network. Hence, the maximum losses of the distribution system are reduced. GA technology is selected as the capacitor problem is more accurately addressed in the power grid. When the search area crosses for an optimal solution, the advantages of GA are multifunctional—that is, when a locally optimal solution is found to an engineering goal, GA adapts its search to find an optimal global solution, subject to predefined search restrictions [ 129 , 130 ]. The article shows the results of the study of the best size and location for a GA-connected system using bays in Saudi Arabia in West–East regions [ 131 , 132 ]. Two formulas are proposed for capacitor positioning: (i) cost balance condenser/loss principle and (ii) total system performance cost estimates, standard analysis and verification of annual benefits, power loss and operational tension results [ 133 , 134 ]. AI is applicable for various aspects of the power system [ 135 ]. Some additional papers on power system application of AI in planning are presented in Table 5 .

Applications of AI in the planning of a power system

Application of AI in various problems of the operation, control and planning of power systems has shown good performance over conventional methods. ABC algorithms function better than other AI techniques in the ELD problem, as shown in a comparison. The ABC algorithm has the highest-quality solutions, consistent convergence and exceptional computing efficiency. Compared with traditional mathematical models, both GA and PSO techniques are superior; the PSO technique is preferable since GA replaces humans. Unlike GA, PSO models alter humans through time and, in the following generation, all people survive. The fate of each person is continually adjusted based on the global ideal point thus far. The swarming effect in PSO allows the population of particles to move collectively in the search space, facilitating a more efficient exploration of the solution space and a faster convergence to optimum search areas compared with GA. PSO is particularly useful for optimization problems with many variables and when the solution space is complex and not well understood. Additionally, PSO requires minimal computational resources compared with GA as it does not require the evaluation and selection of multiple generations of offspring. In UC, with the increasing time, the execution time is also increasing, so the combination of many AI techniques can be progressive, as it can potentially increase the efficiency and accuracy of the optimization problem. The scalability of any heuristic optimization method is a major issue. In the load flow method, the ANFIS and ABC algorithms provide efficient and accurate solutions, and the implementation of ANNs is fast and can handle missing data effectively. In VC, rather than replacing conventional methods, the focus should be on enhancing their capabilities through the integration of AI techniques, modern control theory, fuzzy technology and ANNs, along with adaptive control and expert systems. This approach, which combines current research trends with practical experience, has great potential for practical applications. In load frequency control, the NARMA-L2 ANN network architecture provides the best output after some series of trials and improvements. Under fault conditions, it has been observed. The fuzzy-PID power output works so that the power systems are fast and stable. Due to the specialization, the fuzzy logic in condenser placement is better than other approaches. Fuzzy logic includes the relatively basic technique of calculating the necessary nodes in the distribution system to replace condensers with approximate reasoning. The results of the study indicate that the GA method can provide a globally optimal solution for reactive power optimization, particularly when an ample generation and population size is used. Furthermore, it was found that the use of the UPFC resulted in minimized losses compared with the static var compensator and thyristor-controlled series capacitor. In terms of reactive power planning, a fuzzy membership-based approach is employed in an interconnected power system to identify weak nodes and determine optimal parameter settings for FACTS devices. The feasibility of this strategy is validated through its solutions and compared with other global optimization approaches. The proposed technique is applied to a standard system under high load conditions, resulting in a stable system with reduced losses and cost savings. This approach has the potential to become a novel technology for effectively coordinating FACTS devices with other existing generators. Following the pandemic, it is expected that governments around the world will prioritize energy efficiency in buildings and smart homes. To support this, the development of open-source protocols and unified connectivity solutions is crucial. Smart home systems are focused on maximizing the energy efficiency of major household appliances, thereby contributing to overall energy balance. By constructing sustainable homes integrated with smart technologies and a combination of energy sources, significant cost savings and a reduction in carbon footprints can be achieved. While smart homes are becoming more common, there are still barriers to widespread adoption that researchers need to address. It emphasizes the potential of smart home companies and highlights technical challenges such as device compatibility. The review also discusses the importance of studying consumer attitudes and demands, and mentions the limitations of the survey methodology. Significant scientific results include an algorithm for a modern predictive analytics system, an approach to assess the condition index of the equipment and a method to determine the probability of defects using ML. The study validated the model using data from a hydroelectric power plant, demonstrating its accuracy. Future research can focus on refining the index calculation for equipment with multiple functional units and constructing predictive models for specific equipment classes. ML and data-driven techniques hold great promise in the field of power systems, especially in the context of smart grids. These methods can effectively analyse and process large volumes of data, resulting in improved accuracy and increased operational efficiency. However, some challenges need to be addressed, such as ensuring the quality of the data and interpreting the results in a meaningful way. Accurate forecasting plays a vital role in optimizing grid operations, integrating renewable energy sources and achieving cost-effective power generation. ML plays a vital role in transforming traditional grids into smart grids, improving reliability and safety. It also aids in demand-side management and enhances cybersecurity.

Evaluating energy costs and making improvements can lead to significant energy savings. Smart technologies have the potential to reduce electricity demand and environmental impact. Social acceptance of smart home systems needs to be promoted. Further research can expand sample sizes, include more diverse countries and explore smart meter readings. Future research should emphasize the importance of addressing technical, security and privacy concerns, and call for collaboration between stakeholders to enhance the smart home market. Although the developed approach offers several benefits, there is still an unresolved issue regarding the calculation of the technical condition index for equipment consisting of multiple functional units. Existing methods rely on assigning weights to each unit based on an expert evaluation to determine its importance. Additional research can focus on improving the methods for calculating the technical condition index for different types of power equipment and establishing predictive models to anticipate equipment defects in the event of functional unit failures. Future investigations should prioritize the development of more precise and dependable predictive models for power systems, taking into account the challenges related to data availability and interpretability.

Anshumaan Pathak and Utkarsh Pandey did the critical review. Surajit Mondal and Adesh Kumar supervised and reviewed the manuscript.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

The paper is a study and review article in which no specific data are referred to. No simulation software is used.

Sozontov A , Ivanova M , Gibadullin A. Implementation of artificial intelligence in the electric power industry . E3S Web of Conferences , 2019 , 114 : 01009 .

Google Scholar

Zhao X , Zhang X. Artificial intelligence applications in power system . In: Luo X (ed). Advances in Intelligent Systems Research—2016 2nd International Conference on Artificial Intelligence and Industrial Engineering (AIIE 2016). Dordrecht, Netherlands : Atlantis Press , 2016 , 158 – 161 .

Nath RP , Balaji VN. Artificial intelligence in power systems . IOSR Journal of Computer Engineering (IOSR-JCE) , 2014 . https://jcboseust.ac.in/electrical/images/notes/aitech_ug_ai_reactive_power_control.pdf ( 29 July 2023 , date last accessed).

Das S , Mukherjee M , Mondal S. Detailed energy audit of thermal power plant equipment . World Scientific News , 2015 , 22 : 70 – 90 .

Mondal S , Mondal AK , Sharma A , et al.  . An overview of cleaning and prevention processes for enhancing efficiency of solar photovoltaic panels . Current Sci , 2018 , 115 : 1065 – 1077 .

Wasesa M , Tiara AR , Afrianto MA , et al.  . SARIMA and artificial neural network models for forecasting electricity consumption of a microgrid based educational building . In: 2020 IEEE International Conference on Industrial Engineering and Engineering Management (IEEM), Singapore, 14–17 December 2020 , 210 – 214 .

Jang JS. ANFIS: adaptive-network-based fuzzy inference system . IEEE Trans Syst Man Cybern , 1993 , 23 : 665 – 685 .

Borges AF , Laurindo FJ , Spínola MM , et al.  . The strategic use of artificial intelligence in the digital era: systematic literature review and future research directions . Int J Inf Manage , 2021 , 57 : 102225 .

Guan Y , Chaffart D , Liu G , et al.  . Machine learning in solid heterogeneous catalysis: recent developments, challenges and perspectives . Chem Eng Sci , 2022 , 248 : 117224 .

Bordin C , Skjelbred HI , Kong J , et al.  . Machine learning for hydropower scheduling: state of the art and future research directions . Procedia Comput Sci , 2020 , 176 : 1659 – 1668 .

Wang H , Ricardez-Sandoval LA. Dynamic optimization of a pilot-scale entrained-flow gasifier using artificial recurrent neural networks . Fuel , 2020 , 272 : 117731 .

Isnen M , Kurniawan S , Garcia-Palacios E. A-SEM: an adaptive smart energy management testbed for shiftable loads optimisation in the smart home . Measurement , 2020 , 152 : 107285 .

Bibri SE , Krogstie J. Environmentally data-driven smart sustainable cities: applied innovative solutions for energy efficiency, pollution reduction, and urban metabolism . Energy Informatics , 2020 , 3 : 1 – 59 .

Strielkowski W. Social Impacts of Smart Grids: The Future of Smart Grids and Energy Market Design . Amsterdam, Netherlands : Elsevier , 2019 .

Google Preview

Oliveira L , Mitchell V , May A. Smart home technology—comparing householder expectations at the point of installation with experiences 1 year later . Pers Ubiquitous Comput , 2020 , 24 : 613 – 626 .

Tao M , Zuo J , Liu Z , et al.  . Multi-layer cloud architectural model and ontology-based security service framework for IoT-based smart homes . Future Gener Comput Syst , 2018 , 78 : 1040 – 1051 .

Shcherbatov IA , Turikov GN. Determination of power engineering equipment’s defects in predictive analytic system using machine learning algorithms . J Phys Conf Ser , 2020 , 1683 : 042056 .

Moleda M , Momot A , Mrozek D. Predictive maintenance of boiler feed water pumps using SCADA data . Sensors , 2020 , 20 : 571 .

Ren C , Xu Y. A fully data-driven method based on generative adversarial networks for power system dynamic security assessment with missing data . IEEE Trans Power Syst , 2019 , 34 : 5044 – 5052 .

Yu J , Guo Y , Sun H. Testbeds for integrated transmission and distribution networks: generation methodology and benchmarks . CSEE J Power Energy Syst , 2020 , 6 : 518 – 527 .

Gonen T. Electric Power Distribution Engineering . Boca Raton, FL, USA : CRC Press , 2015 .

Wood AJ , Wollenberg BF , Sheblé GB. Power Generation, Operation, and Control . Hoboken, NJ, USA : John Wiley & Sons , 2013 .

Kumar N , Mishra VM , Kumar A. Smart grid and nuclear power plant security by integrating cryptographic hardware chip . Nuclear Engineering and Technology , 2021 , 53 : 3327 – 3334 .

Kumar N , Mishra VM , Kumar A. Smart grid security by embedding s-box advanced encryption standard . Intelligent Automation &. Soft Comput , 2022 , 34 : 623 – 638 .

Xu G , Wang Z. Power system load flow distribution research based on adaptive neuro-fuzzy inference systems . In: 2012 Spring Congress on Engineering and Technology, Xi’an, China, 27–30 May 2012 , 1 – 4 .

Karaboga D , Akay B. A comparative study of artificial bee colony algorithm . Appl Math Comput , 2009 , 214 : 108 – 132 .

Nualhong D , Chusanapiputt S , Phomvuttisarn S , et al.  . Diversity control approach to ant colony optimization for unit commitment problem . In: 2004 IEEE Region 10 Conference TENCON 2004, Chiang Mai, Thailand, 24 November 2004 , 488 – 491 .

Ho SL , Yang S , Ni G , et al.  . A particle swarm optimization-based method for multiobjective design optimizations . IEEE Trans Magn , 2005 , 41 : 1756 – 1759 .

Ratnaweera A , Halgamuge SK , Watson HC. Self-organizing hierarchical particle swarm optimizer with time-varying acceleration coefficients . IEEE Trans Evol Comput , 2004 , 8 : 240 – 255 .

Kumar N , Nangia U , Sahay KB. Economic load dispatch using improved particle swarm optimization algorithms . In: 2014 6th IEEE Power India International Conference (PIICON), Delhi, India, 5–7 December 2014 , 1 – 6 .

Panta S , Premrudeepreechacharn S. Economic dispatch for power generation using artificial neural network ICPE’07 conference in Daegu, Korea . In: 2007 7th International Conference on Power Electronics, Daegu, South Korea, 22–26 October 2007 , 558 – 562 .

Rahmat NA , Musirin I. Differential evolution ant colony optimization (DEACO) technique in solving economic load dispatch problem . In: 2012 IEEE International Power Engineering and Optimization Conference, Melaka, Malaysia, 6–7 June 2012 , 263 – 268 .

Saxena A , Pandey SN , Srivastava L. Congestion management in open access—a review . International Journal of Science, Engineering and Technology Research , 2013 , 2 : 922 – 930 .

Emami H , Sadri JA. Congestion management of transmission lines in the market environment . International Research Journal of Applied and Basic Sciences , 2012 , 3 : 2572 – 2580 .

Ghosh K , Pandey U , Pathak A , et al.  . Simulation of density based traffic control system using Proteus 7.1 professional . In: Mathur G , Bundele M , Tripathi A , Paprzycki M (eds). Proceedings of 3rd International Conference on Artificial Intelligence: Advances and Applications: ICAIAA 2022. Singapore: Springer Nature Singapore , 2023 , 493 – 504 .

Deb S , Goswami AK. Congestion management by generator rescheduling using artificial bee colony optimization technique . In: 2012 Annual IEEE India Conference (INDICON), Kochi, India, 7–9 December 2012 , 909 – 914 .

Choudekar P , Sinha SK , Siddiqui A. Transmission line efficiency improvement and congestion management under critical contingency condition by optimal placement of TCSC . In: 2016 7th India International Conference on Power Electronics (IICPE), Patiala, India, 17–19 November 2016 , 1 – 6 .

Wang X , McDonald JR. Modern Power System Planning . London, UK : McGraw-Hill , 1994 , 247 – 307 .

Dahal K , McDonald J. A review of generator maintenance scheduling using artificial intelligence techniques . In: Proceedings of the 32nd Universities Power Engineering Conference (UPEC ’97), Manchester, UK, 10–12 September 1997 , 787 – 790 .

Koay CA , Srinivasan D. Particle swarm optimization-based approach for generator maintenance scheduling . In: Proceedings of the 2003 IEEE Swarm Intelligence Symposium. SIS’03 (Cat. No. 03EX706), Indianapolis, IN, USA, 26 April 2003 , 167 – 173 .

Johnson RC , Happ HH , Wright WJ. Large scale hydro-thermal unit commitment: method and results . IEEE Trans Power Appar Syst , 1971 , PAS-90 : 1373 – 1384 .

Thum YM. Hierarchical linear models for multivariate outcomes . Journal of Educational and Behavioral Statistics , 1997 , 22 : 77 – 108 .

Reyes DM , de Souza RM , de Oliveira AL. A three-stage approach for modeling multiple time series applied to symbolic quartile data . Expert Syst Appl , 2022 , 187 : 115884 .

Almaghrebi A , Aljuheshi F , Rafaie M , et al.  . Data-driven charging demand prediction at public charging stations using supervised machine learning regression methods . Energies , 2020 , 13 : 4231 .

Huang P , Copertaro B , Zhang X , et al.  . A review of data centers as prosumers in district energy systems: renewable energy integration and waste heat reuse for district heating . Appl Energy , 2020 , 258 : 114109 .

Sun L , You F. Machine learning and data-driven techniques for the control of smart power generation systems: an uncertainty handling perspective . Engineering , 2021 , 7 : 1239 – 1247 .

Lisin E , Shuvalova D , Volkova I , et al.  . Sustainable development of regional power systems and the consumption of electric energy . Sustainability , 2018 , 10 : 1111 .

Prakash S , Jain J , Hasnat S , et al.  . Economic load dispatch with valve point loading effect using optimization techniques . In: Bansal P , Tushir M , Balas VE , Srivastava R (eds). Proceedings of International Conference on Artificial Intelligence and Applications: ICAIAA 2020, Singapore: Springer , 2021 , 407 – 416 .

Chowdhury AK , Mondal S , Mukherjee M , Biswas PK. Mega watt security assessment of power systems . International Letters of Chemistry, Physics and Astronomy, 2015 , 58 : 9 – 15 .

Dommel HW , Tinney WF. Optimal power flow solutions . IEEE Transactions on Power Apparatus and Systems, 1968 , PAS-87 : 1866 – 1876 .

Wakiru JM , Pintelon L , Muchiri P , et al.  . A comparative analysis of maintenance strategies and data application in asset performance management for both developed and developing countries . International Journal of Quality & Reliability Management , 2022 , 39 : 961 – 983 .

Muñoz-Delgado G , Contreras J , Arroyo JM , et al.  . Integrated transmission and distribution system expansion planning under uncertainty . IEEE Trans Smart Grid , 2021 , 12 : 4113 – 4125 .

Chowdhury AK , Mondal S , Alam SM , et al.  . Voltage security assessment of power system . World Scientific News , 2015 , 21 : 36 – 50 .

Holen AT , Botnen A , Stoa P , et al.  . Coupling between knowledge-based and algorithmic methods . Proc IEEE , 1992 , 80 : 745 – 757 .

Short MJ , Hui KC , Ekwue AO , et al.  . Applications of artificial neural networks for NGC voltage collapse monitoring . International Conference On Large High Voltage Electric Systems, 1994 , 2 : 38 – 205 .

Haida T , Akimoto T. Voltage optimization using genetic algorithm . In: Proc. 3rd Symposium on Expert System Applications to Power Systems, Tokyo, Japan, 1991 , 375 – 380 .

Denny FI , Dismukes DE. Power System Operations and Electricity Markets . Boca Raton, FL, USA : CRC Press , 2017 .

Masiala M , Ghribi M , Kaddouri A. An adaptive fuzzy controller gain scheduling for power system load-frequency control . In: 2004 IEEE International Conference on Industrial Technology, IEEE ICIT’04, Vol. 3 , Hammamet, Tunisia, 8–10 December 2004 , 1515 – 1520 .

Alquthami T , Butt SE , Tahir MF , et al.  . Short-term optimal scheduling of hydro-thermal power plants using artificial bee colony algorithm . Energy Rep , 2020 , 6 : 984 – 992 .

Sahay KB , Sonkar A , Kumar A. Economic load dispatch using genetic algorithm optimization technique . In: 2018 International Conference and Utility Exhibition on Green Energy for Sustainable Development (ICUE), Phuket, Thailand, 24–26 October 2018 , 1 – 5 .

Mishra SK , Mishra SK. A comparative study of solution of economic load dispatch problem in power systems in the environmental perspective . Procedia Comput Sci , 2015 , 48 : 96 – 100 .

Dixit GP , Dubey HM , Pandit M , et al.  . Artificial bee colony optimization for combined economic load and emission dispatch . In: International Conference on Sustainable Energy and Intelligent Systems (SEISCON 2011), Chennai, India, 20–22 July 2011 , 340 – 345 .

Daniel L , Chaturvedi KT , Kolhe ML. Dynamic economic load dispatch using Levenberg Marquardt algorithm . Energy Procedia , 2018 , 144 : 95 – 103 .

Ruiz-Abellón MC , Fernández-Jiménez LA , Guillamón A , et al.  . Integration of demand response and short-term forecasting for the management of prosumers’ demand and generation . Energies , 2020 , 13 : 11 .

Ali SS , Choi BJ. State-of-the-art artificial intelligence techniques for distributed smart grids: a review . Electronics , 2020 , 9 : 1030 .

Fu J , Nunez A , De Schutter B. A short-term preventive maintenance scheduling method for distribution networks with distributed generators and batteries . IEEE Trans Power Syst , 2020 , 36 : 2516 – 2531 .

Esmaili M , Shayanfar HA , Moslemi R. Locating series FACTS devices for multi-objective congestion management improving voltage and transient stability . Eur J Oper Res , 2014 , 236 : 763 – 773 .

Suresh K , Kumarappan N. Hybrid improved binary particle swarm optimization approach for generation maintenance scheduling problem . Swarm Evol Comput , 2013 , 9 : 69 – 89 .

Lakshminarayanan S , Kaur D. Optimal maintenance scheduling of generator units using discrete integer cuckoo search optimization algorithm . Swarm Evol Comput , 2018 , 42 : 89 – 98 .

Scalabrini Sampaio G , Vallim Filho ARDA , Santos da Silva L , et al.  . Prediction of motor failure time using an artificial neural network . Sensors , 2019 , 19 : 4342 .

Fikri M , Cheddadi B , Sabri O. Predicting Moroccan real network’s power flow employing the artificial neural networks . In: 2019 Third International Conference on Intelligent Computing in Data Sciences (ICDS), Marrakech, Morocco, 28–30 October 2019 , 1 – 6 .

Rahul J , Sharma Y , Birla D. A new attempt to optimize optimal power flow-based transmission losses using genetic algorithm . In: 2012 Fourth International Conference on Computational Intelligence and Communication Networks, Mathura, India, 3–5 November 2012 , 566 – 570 .

Nakawiro W , Erlich I. A combined GA-ANN strategy for solving optimal power flow with voltage security constraint . In: 2009 Asia-Pacific Power and Energy Engineering Conference, Wuhan, China, 27–31 March 2009 , 1 – 4 .

Sumpavakup C , Srikun I , Chusanapiputt S. A solution to the optimal power flow using artificial bee colony algorithm . In: 2010 International Conference on Power System Technology, Zhejiang, China, 24–28 October 2010 , 1 – 5 .

Abdellah D , Djamel L. Power flow analysis using adaptive neuro-fuzzy inference systems . In: 2015 3rd International Renewable and Sustainable Energy Conference (IRSEC), Marrakech, Morocco, 10–13 December 2015 , 1 – 5 .

Nemati M , Braun M , Tenbohlen S. Optimization of unit commitment and economic dispatch in microgrids based on genetic algorithm and mixed integer linear programming . Appl Energy , 2018 , 210 : 944 – 963 .

Alshareef A. An application of artificial intelligent optimization techniques to dynamic unit commitment for the western area of Saudi Arabia . In: 2011 Third International Conference on Computational Intelligence, Communication Systems and Networks, Bali, Indonesia, 26–28 July 2011 , 17 – 21 .

Arora I , Kaur M. Unit commitment scheduling by employing artificial neural network-based load forecasting . In: 2016 7th India International Conference on Power Electronics (IICPE), Patiala, India, 17–19 November 2016 , 1 – 6 .

Liu Z , Li N , Zhang C. Unit commitment scheduling using a hybrid ANN and Lagrangian relaxation method . In: 2008 International Conference on Multimedia and Ubiquitous Engineering (mue 2008), Busan, South Korea, 24–25 April 2008 , 481 – 484 .

Kumar VS , Mohan MR. Solution to security constrained unit commitment problem using genetic algorithm . International Journal of Electrical Power & Energy Systems , 2010 , 32 : 117 – 125 .

Prakash S , Sinha SK. Application of artificial intelligence in load frequency control of interconnected power system . International Journal of Engineering, Science and Technology , 2011 , 3 : 264 – 275 .

Moshtagh J , Rafinia A. A new approach to high impedance fault location in three-phase underground distribution system using combination of fuzzy logic & wavelet analysis . In: 2012 11th International Conference on Environment and Electrical Engineering, Venice, Italy, 15–18 May 2012 , 90 – 97 .

Santoso NT , Tan OT. Neural-net based realtime control of capacitors installed on distribution systems . IEEE Trans. Power Delivery , 1990 , 5 : 266 – 272 .

Ng HN , Salama MMA , Chikhani AY. Capacitor allocation by approximate reasoning: fuzzy capacitor placement . IEEE Trans Power Deliv , 2000 , 15 : 393 – 398 .

Zhang H , Zhang L , Meng F. Reactive power optimization based on genetic algorithm . In: POWERCON’98. International Conference on Power System Technology. Proceedings, Vol. 2 , Beijing, China, 18–21 August 1998 , 1448 – 1453 .

Mamandur KRC , Chenoweth RD. Optimal control of reactive power flow for improvements in voltage profiles and for real power loss minimization . IEEE Transactions on Power Apparatus and Systems, 1981 , PAS-100 : 3185 – 3194 .

Kothakotla V , Kumar B. Analysis and design of robust PID controller for grid voltage control of islanded microgrid using genetic algorithm . In: 2021 5th International Conference on Intelligent Computing and Control Systems (ICICCS), Madurai, India, 6–8 May 2021 , 719 – 726 .

Wang S , Duan J , Shi D , et al.  . A data-driven multi-agent autonomous voltage control framework using deep reinforcement learning . IEEE Trans Power Syst , 2020 , 35 : 4644 – 4654 .

Zidani Y , Zouggar S , Elbacha A. Steady-state analysis and voltage control of the self-excited induction generator using artificial neural network and an active filter . Iranian Journal of Science and Technology, Transactions of Electrical Engineering , 2018 , 42 : 41 – 48 .

Sumathi S. Artificial neural network application for voltage control and power flow control in power systems with UPFC . In: 2015 International Conference on Emerging Research in Electronics, Computer Science and Technology (ICERECT), Mandya, India, 17–19 December 2015 , 403 – 407 .

Kanata S , Sianipar GH , Maulidevi NU. Optimization of reactive power and voltage control in power system using hybrid artificial neural network and particle swarm optimization . In: 2018 2nd International Conference on Applied Electromagnetic Technology (AEMT), Lombok, Indonesia, 9–12 April 2018 , 67 – 72 .

Abdalla OH , Ghany AA , Fayek HH. Coordinated PID secondary voltage control of a power system based on genetic algorithm . In: 2016 Eighteenth International Middle East Power Systems Conference (MEPCON), Cairo, Egypt, 27–29 December 2016 , 214 – 219 .

Chung IY , Liu W , Cartes DA. Control parameter optimization for a microgrid system using particle swarm optimization . In: 2008 IEEE International Conference on Sustainable Energy Technologies, Singapore, 24–27 November 2008 , 837 – 842 .

Yousuf H , Zainal AY , Alshurideh M , et al.  . Artificial intelligence models in power system analysis . In: Hassanien AE , Bhatnagar R , Darwish A (eds). Artificial Intelligence for Sustainable Development: Theory, Practice and Future Applications . Cham, Switzerland : Springer , 2021 , 231 – 242 .

Aakula JL , Khanduri A , Sharma A. Determining reactive power levels to improve bus voltages using PSO . In: 2020 IEEE 17th India Council International Conference (INDICON), New Delhi, India, 10–13 December 2020 , 1 – 7 .

Karthikeyan R , Pasam S , Sudheer S , et al.  . Fuzzy fractional order PID based parallel cascade control system . In: Thampi S , Abraham A , Pal S , Rodriguez J (eds). Recent Advances in Intelligent Informatics. Advances in Intelligent Systems and Computing , Vol. 235 . Cham, Switzerland : Springer , 2014 , 293 – 302 .

Sallama A , Abbod M , Taylor G. Supervisory power system stability control using neuro-fuzzy system and particle swarm optimization algorithm . In: 2014 49th International Universities Power Engineering Conference (UPEC), Cluj-Napoca, Romania, 2–5 September 2014 , 1 – 6 .

Chen B , Li W , Yu P , et al.  . Planning approach for cross-regional optical transmission network supporting wide-area stability control services in power system . In: 2018 International Conference on Network Infrastructure and Digital Content (IC-NIDC), Guiyang, China, 22–24 August 2018 , 105 – 109 .

Torkzadeh R , NasrAzadani H , Aliabad AD , et al.  . A genetic algorithm optimized fuzzy logic controller for UPFC in order to damp of low frequency oscillations in power systems . In: 2014 22nd Iranian Conference on Electrical Engineering (ICEE), Tehran, Iran, 20–22 May 2014 , 706 – 712 .

Dutta S , Singh SP. Optimal rescheduling of generators for congestion management based on particle swarm optimization . IEEE Trans Power Syst , 2008 , 23 : 1560 – 1569 .

Nam S , Hur J. Probabilistic forecasting model of solar power outputs based on the naive Bayes classifier and kriging models . Energies , 2018 , 11 : 2982 .

Miraftabzadeh SM , Longo M , Foiadelli F , et al.  . Advances in the application of machine learning techniques for power system analytics: a survey . Energies , 2021 , 14 : 4776 .

Safari A , Babaei F , Farrokhifar M. A load frequency control using a PSO-based ANN for micro-grids in the presence of electric vehicles . Int J Ambient Energy , 2021 , 42 : 688 – 700 .

Joshi M , Sharma G , Davidson IE. Load frequency control of hydro electric system using application of fuzzy with particle swarm optimization algorithm . In: 2020 International Conference on Artificial Intelligence, Big Data, Computing and Data Communication Systems (icABCD), Durban, South Africa, 6–7 August 2020 , 1 – 6 .

Nguyen GN , Jagatheesan K , Ashour AS , et al.  . Ant colony optimization-based load frequency control of multi-area interconnected thermal power system with governor dead-band nonlinearity . In: Yang XS , Nagar AK , Joshi A (eds). Smart Trends in Systems, Security and Sustainability . Singapore : Springer , 2018 , 157 – 167 .

Balamurugan CR. Three area power system load frequency control using fuzzy logic controller . International Journal of Applied Power Engineering (IJAPE) , 2018 , 7 : 18 – 26 .

Otani T , Tanabe R , Koyanagi Y , et al.  . Cooperative load frequency control of generator and battery using a recurrent neural network . In: TENCON 2017–2017 IEEE Region 10 Conference, Penang, Malaysia, 5–8 November 2017 , 918 – 923 .

Kumar D , Mathur HD , Bhanot S , et al.  . Forecasting of solar and wind power using LSTM RNN for load frequency control in isolated microgrid . Int J Modelling Simul , 2021 , 41 : 311 – 323 .

Arora K , Kumar A , Kamboj VK , et al.  . Optimization methodologies and testing on standard benchmark functions of load frequency control for interconnected multi area power system in smart grids . Mathematics , 2020 , 8 : 980 .

Vadivelu KR , Marutheswar GV. Artificial intelligence technique based reactive power planning incorporating FACTS controllers in real time power transmission system . In: 2014 IEEE 2nd International Conference on Electrical Energy Systems (ICEES), Chennai, India, 7–9 January 2014 , 26 – 31 .

Khan MA , Algarni F. A healthcare monitoring system for the diagnosis of heart disease in the IoMT cloud environment using MSSO-ANFIS . IEEE Access , 2020 , 8 : 122259 – 122269 .

Liu M , Dong M , Wu C. A new ANFIS for parameter prediction with numeric and categorical inputs . IEEE Trans Autom Sci Eng , 2010 , 7 : 645 – 653 .

Ghadiri A , Haghifam MR , Larimi SMM. Comprehensive approach for hybrid AC/DC distribution network planning using genetic algorithm . IET Generation, Transmission & Distribution , 2017 , 11 : 3892 – 3902 .

Jain SK , Bhargava A , Pal RK. Three area power system load frequency control using fuzzy logic controller . In: 2015 International Conference on Computer, Communication and Control (IC4), Indore, India, 10–12 September 2015 , 1 – 6 .

Son YS , Kim HJ , Kim JT. A video-quality control scheme using ANFIS architecture in a DASH environment . Journal of Broadcast Engineering , 2018 , 23 : 104 – 114 .

Kannadasan K , Edla DR , Yadav MH , et al.  . Intelligent-ANFIS model for predicting measurement of surface roughness and geometric tolerances in three-axis CNC milling . IEEE Trans Instrum Meas , 2020 , 69 : 7683 – 7694 .

Penghui L , Ewees AA , Beyaztas BH , et al.  . Metaheuristic optimization algorithms hybridized with artificial intelligence model for soil temperature prediction: novel model . IEEE Access , 2020 , 8 : 51884 – 51904 .

Tawfeek TS , Ahmed AH , Hasan S. Analytical and particle swarm optimization algorithms for optimal allocation of four different distributed generation types in radial distribution networks . Energy Procedia , 2018 , 153 : 86 – 94 .

Yeom CU , Kwak KC. Adaptive neuro-fuzzy inference system predictor with an incremental tree structure based on a context-based fuzzy clustering approach . Applied Sciences , 2020 , 10 : 8495 .

Krasopoulos CT , Beniakar ME , Kladas AG. Multicriteria PM motor design based on ANFIS evaluation of EV driving cycle efficiency . IEEE Trans Transp Electrif , 2018 , 42 : 525 – 535 .

El-Hasnony IM , Barakat SI , Mostafa . Optimized ANFIS model using hybrid metaheuristic algorithms for Parkinson’s disease prediction in IoT environment . IEEE Access , 2020 , 8 : 119252 – 119270 .

Morshedizadeh M , Kordestani M , Carriveau R , et al.  . Power production prediction of wind turbines using a fusion of MLP and ANFIS networks . IET Renew Power Gener , 2018 , 12 : 1025 – 1033 .

Khosravi A , Nahavandi S , Creighton D. Prediction interval construction and optimization for adaptive neurofuzzy inference systems . IEEE Trans Fuzzy Syst , 2011 , 19 : 983 – 988 .

Elbaz K , Shen SL , Sun WJ , et al.  . Prediction model of shield performance during tunneling via incorporating improved particle swarm optimization into ANFIS . IEEE Access , 2020 , 8 : 39659 – 39671 .

Dovžan D , Škrjanc I. Fuzzy space partitioning based on hyperplanes defined by eigenvectors for Takagi-Sugeno fuzzy model identification . IEEE Trans Ind Electron , 2019 , 67 : 5144 – 5153 .

Castiello C , Fanelli AM , Lucarelli M , et al.  . Interpretable fuzzy partitioning of classified data with variable granularity . Appl Soft Comput , 2019 , 74 : 567 – 582 .

Alexandridis A , Chondrodima E , Sarimveis H. Radial basis function network training using a nonsymmetric partition of the input space and particle swarm optimization . IEEE Trans Neural Networks Learn Syst , 2012 , 24 : 219 – 230 .

Verstraete J. The spatial disaggregation problem: simulating reasoning using a fuzzy inference system . IEEE Trans Fuzzy Syst , 2016 , 25 : 627 – 641 .

Olamaei J , Moradi M , Kaboodi T. A new adaptive modified firefly algorithm to solve optimal capacitor placement problem . In: 18th Electric Power Distribution Conference, Kermanshah, Iran, 30 April–1 May 2013 , 1 – 6 .

Lee JS , Teng CL. An enhanced hierarchical clustering approach for mobile sensor networks using fuzzy inference systems . IEEE Internet Things J , 2017 , 4 : 1095 – 1103 .

Su ZG , Denoeux T. BPEC. Belief-peaks evidential clustering . IEEE Trans Fuzzy Syst , 2018 , 27 : 111 – 123 .

Xu P , Deng Z , Cui C , et al.  . Concise fuzzy system modeling integrating soft subspace clustering and sparse learning . IEEE Trans Fuzzy Syst , 2019 , 27 : 2176 – 2189 .

Sujil A , Kumar R , Bansal RC. FCM clustering-ANFIS-based PV and wind generation forecasting agent for energy management in a smart microgrid . The Journal of Engineering , 2019 , 2019 : 4852 – 4857 .

Neamatollahi P , Naghibzadeh M , Abrishami S. Fuzzy-based clustering-task scheduling for lifetime enhancement in wireless sensor networks . IEEE Sens J , 2017 , 17 : 6837 – 6844 .

Wang Z , Xu Y. A golden section-based double population genetic algorithm applied to reactive power optimization . IOP Conference Series: Earth and Environmental Science, 2021 , 645 : 012074 .

Kahouli O , Alsaif H , Bouteraa Y , et al.  . Power system reconfiguration in distribution network for improving reliability using genetic algorithm and particle swarm optimization . Applied Sciences , 2021 , 11 : 3092 .

Žarković SD , Stanković S , Shayesteh E , et al.  . Reliability improvement of distribution system through distribution system planning: MILP vs. GA . In: 2019 IEEE Milan PowerTech, Milan, Italy, 23–27 June 2019, 1 – 6 .

Ahmetovic H , Saric M , Hivziefendic J. Reliability based power distribution network planning using fuzzy logic . Advances in Electrical and Electronic Engineering , 2021 , 19 : 123 – 133 .

Suresh M , Sirish TS , Subhashini TV , et al.  . Load flow analysis of distribution system using artificial neural networks . In: Satapathy SC , Bhateja V , Udgata SK , Pattnaik PK (eds). Proceedings of the 5th International Conference on Frontiers in Intelligent Computing: Theory and Applications FICTA 2016, Vol. 1 . Springer , Singapore , 2017 , 515 – 524 .

Kumari M , Ranjan R , Singh VR. Optimal power distribution planning using improved particle swarm optimization . International Journal of Intelligent Systems and Applications in Engineering , 2018 , 3 : 170 – 177 .

Saha S , Mukherjee V. A novel multi-objective modified symbiotic organisms search algorithm for optimal allocation of distributed generation in radial distribution system . Neural Comput Appl , 2021 , 33 : 1751 – 1771 .

Hosseini MM , Parvania M. Artificial intelligence for resilience enhancement of power distribution systems . The Electricity Journal , 2021 , 34 : 106880 .

Lytras MD , Chui KT. The recent development of artificial intelligence for smart and sustainable energy systems and applications . Energies , 2019 , 12 : 3108 .

Gandhar S , Ohri J , Singh M. Dynamic reactive power optimization of hybrid micro grid in islanded mode using fuzzy tuned UPFC . J Inf Optim Sci , 2020 , 41 : 305 – 315 .

Tang Y , Hu W , Cao D , et al.  . Artificial intelligence-aided minimum reactive power control for the DAB converter based on harmonic analysis method . IEEE Trans Power Electron , 2021 , 36 : 9704 – 9710 .

Harrye YA , Ahmed KH , Aboushady AA. Reactive power minimization of dual active bridge DC/DC converter with triple phase shift control using neural network . In: 2014 International Conference on Renewable Energy Research and Application (ICRERA), Milwaukee, WI, USA, 19–22 October 2014 , 566 – 571 .

Wang Y , Li Y. A hybrid ant colony optimization algorithm for dynamic optimization of vehicle routing problem with time windows . Applied Intelligence , 2021 , 51 : 476 – 491 .

Sharma NK , Babu DS , Choube SC. Application of particle swarm optimization technique for reactive power optimization . In: IEEE International Conference on Advances in Engineering, Science and Management (ICAESM-2012), Nagapattinam, India, 30–31 March 2020 , 88 – 93 .

Bhattacharyya B , Gupta VK. Fuzzy based evolutionary algorithm for reactive power optimization with FACTS devices . International Journal of Electrical Power & Energy Systems , 2014 , 61 : 39 – 47 .

Bharti D. Multi-point optimal placement of shunt capacitor in radial distribution network: a comparison . In: 2020 International Conference on Emerging Frontiers in Electrical and Electronic Technologies (ICEFEET), Patna, India, 10–11 July 2020 , 1 – 6 .

Roy PK , Sultana S. Optimal reconfiguration of capacitor based radial distribution system using chaotic quasi oppositional chemical reaction optimization . Microsyst Technol , 2022 , 28 : 499 – 511 .

Pimentel Filho MC , De Lacerda EGM , Junior MM. Capacitor placement using ant colony optimization and gradient . In: 2009 15th International Conference on Intelligent System Applications to Power Systems, Curitiba, Brazil, 8–12 November 2009 , 1 – 4 .

Isac SJ , Kumar KS , Kumar PV. Optimal capacitor placement in radial distribution system to minimize the loss using fuzzy logic control . In: International Conference on Smart Structures and Systems—ICSSS; 13 , Chennai, India, 28–29 March 2013 , 33 – 40 .

Reddy MD , Reddy VV. Capacitor placement using fuzzy and particle swarm optimization method for maximum annual savings . ARPN Journal of Engineering and Applied Sciences , 2008 , 3 : 25 – 30 .

Shwehdi MH , Mohamed SR , Devaraj D. Optimal capacitor placement on West–East inter-tie in Saudi Arabia using genetic algorithm . Computers & Electrical Engineering , 2018 , 68 : 156 – 169 .

Mahdavian M , Kafi MH , Movahedi A , et al.  . Improve performance in electrical power distribution system by optimal capacitor placement using genetic algorithm . In: 2017 14th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications, and Information Technology (ECTI-CON), Phuket, Thailand, 27–30 June 2017 , 749 – 752 .

Email alerts

Citing articles via.

• Advertising and Corporate Services

Affiliations

• Online ISSN 2515-396X
• Print ISSN 2515-4230
• Copyright © 2024 National Institute of Clean-and-Low-Carbon Energy
• About Oxford Academic
• Publish journals with us
• University press partners
• What we publish
• New features  
• Open access
• Institutional account management
• Rights and permissions
• Get help with access
• Accessibility
• Advertising
• Media enquiries
• Oxford University Press
• Oxford Languages
• University of Oxford

Oxford University Press is a department of the University of Oxford. It furthers the University's objective of excellence in research, scholarship, and education by publishing worldwide

• Copyright © 2024 Oxford University Press
• Cookie settings
• Cookie policy
• Privacy policy
• Legal notice

This Feature Is Available To Subscribers Only

Sign In or Create an Account

This PDF is available to Subscribers Only

For full access to this pdf, sign in to an existing account, or purchase an annual subscription.

Improving the performance of power system protection using wide area monitoring systems

• Special Issue on Wide Area Monitoring, Protection and Control in Smart Grid
• Open access
• Published: 13 July 2016
• Volume 4 , pages 319–331, ( 2016 )

Cite this article

You have full access to this open access article

• Arun G. PHADKE 1 ,
• Peter WALL 2 ,
• Lei DING 3 &
• Vladimir TERZIJA 2  

18k Accesses

63 Citations

1 Altmetric

Explore all metrics

Wide area monitoring (WAM) offers many opportunities to improve the performance of power system protection. This paper presents some of these opportunities and the motivation for their development. This methods include monitoring the suitability of relay characteristics, supervisory control of backup protection, more adaptive and intelligent system protection and the creation of novel system integrity protection scheme. The speed of response required for primary protection means that the role WAM in enhancing protection is limited to backup and system protection. The opportunities offered by WAM for enhancing protection are attractive because of the emerging challenges faced by the modern power system protection. The increasingly variable operating conditions of power systems are making it ever more difficult to select relay characteristics that will be a suitable compromise for all loading conditions and contingencies. The maloperation of relays has contributed to the inception and evolution of 70 % of blackouts, thus the supervision of the backup protection may prove a valuable tool for preventing or limiting the scale of blackouts. The increasing interconnection and complexity of modern power systems has made them more vulnerable to wide area disturbances and this has contributed to several recent blackouts. The proper management of these wide area disturbances is beyond the scope of most of the existing protection and new, adaptive system integrity protection schemes are needed to protect power system security.

Similar content being viewed by others

Introduction to Condition Monitoring of Wide Area Monitoring System

A Comprehensive Review on Wide-Area Protection, Control and Monitoring Systems

Wide Area Monitoring System

Avoid common mistakes on your manuscript.

1 Introduction

Wide area monitoring (WAM) is one of the most significant new developments in modern power systems. Through developments in synchronized measurement technology and the creation of phasor measurement units (PMUs) [ 1 ], WAM is able to offer a real time view of the dynamic behavior of a power system that updates once per cycle. This information has proven an invaluable resource for creating new applications that can benefit power system protection and control [ 2 – 6 ].

Recent blackout reports have identified that failings in protection systems have contributed to several recent blackouts [ 7 , 8 ]. Therefore, the role that WAM may be able to play in enhancing power system protection has become an area of great interest.

The speed of response required for primary protection is too high for wide area measurements to play a role. Furthermore, the need for wide area measurements as part of primary protection is limited, as it protects a specific element of the power system. However, aspects of power system protection that have lower requirements in terms of the speed of response (e.g. backup protection) and are less selective can be improved by using wide area measurements to supervise their behavior. Furthermore, wide area measurements can be used as the basis for creating adaptive system protection, novel system integrity protection schemes, or even entirely new protection concepts (e.g. real time adaptation of the balance between security and dependability).

Wide area measurements alone are not sufficient to realize these potential enhancements. The introduction of digital relays has provided an unprecedented level of computational power in the substation and this has vastly increased the scope of the functions that can be delivered by any protection system. This enhanced capability is already leading to an increasing amount of intelligence and decision making moving from the control center to the substation [ 9 ] and the new protection concepts discussed here are an extension of this.

However, in addition to this increased computational power and the availability of wide area measurements, a key requirement for any wide area application is a suitable communication infrastructure to support it.

The communication needs of different WAP concepts can vary drastically [ 10 ]. Some may require measurements to be streamed from multiple locations at a rate of once per cycle (e.g. intelligent controlled islanding [ 11 ]) while others may only require binary signals to be streamed at lower rates (e.g. supervision of backup protection [ 10 ]).

Furthermore, the requirements imposed on the communication infrastructure extend beyond bandwidth. The latency and jitter may need to be low, so that a reliable, high speed of response is provided, and ensuring cyber security will be very important to prevent WAP from being exploited by malicious third parties that seek to attack the power system. Therefore, proper evaluation of the communication needs should form an essential aspect of the design of any wide area protection scheme [ 12 ].

The increasing relevance of WAP is driven by the changing nature of power systems. The three main drivers are: ① The wider range of possible operating conditions, due to the changing generation mix and the introduction of demand side participation; ② the increased interconnection of power systems, larger infeeds from neighboring systems and the reduction in operating margins due to economic pressures; and ③ the increasing complexity and diversity of transmission technology and control (e.g. HVDC, thyristor controlled series compensation, increasing interconnection).

These changes are making it increasingly difficult to select protection settings that will be an appropriate compromise for all credible system conditions and contingencies. Furthermore, modern power systems are more vulnerable to wide area disturbances. Wide area disturbances require a coordinated wide area response across system boundaries that is tailored to the needs of the entire system, not inaccurate, inconsistent local responses that are delivered based on the local observations of each system.

It has been reported [ 10 ] that 70 % of wide area disturbances involved relay maloperation during their initiation or evolution. These maloperations can be attributed to either poor relay settings or hidden failures in the protection system. The role of relay maloperation in wide area disturbances must be taken as a significant source of concern, as wide area disturbances have played a key role in several recent blackouts [ 7 , 8 ] and the management of these wide area disturbances is beyond the scope of most of the existing protection [ 13 ].

These factors have motivated the development of new protection concepts that are supported by WAM. The varied nature of the challenges facing protection has meant that these new concepts cover a broad range of complexity and ambition. Examples include novel system integrity protection schemes (SIPS) that can deploy a wide range of far reaching actions to prevent a cascading failure, adaptive system protection (e.g. adaptive under frequency load shedding), supervisory schemes that improve the security of existing backup protection, and methods that do not change the behavior of system protection but do enhance our understanding of it (e.g. alarming system operators to the risk of false penetration of relay characteristics). Recent work has begun to focus not only on developing new concepts but also on the practical realization of these concepts, e.g. work has addressed the use of the IEEE 1588 std for substation synchronization as part of the Guizhou-Duyun WAP project in Guizhou province China [ 14 ].

This paper describes a number of the proposed concepts and how they can help to address several significant threats to the proper performance of power system protection, including:

The role of cascade failures and wide area disturbances in power system blackouts

Ensuring the security of backup relays in the more complex operating conditions of modern power systems

Limiting the impact of hidden failures that are revealed under stressed conditions

The adaptation of system protection actions to the true system state

Wide area protection of distribution systems

The paper is structured as follows. Section  2 introduces some basic aspects of WAM and PMUs. Section  3 provides an overview of power system protection and the threats that it faces. Section  4 describes a section of the new protection concepts that are being developed. Finally, Section  5 provides some concluding remarks.

2 Wide area monitoring

WAM collects measurements from remote locations across the power system and combines them in real time into a single snapshot of the power system for a given time. Synchronized measurement technology (SMT) is an essential component of WAM, as it allows the measurements to be accurately time stamped, primarily using timing signals from GPS. These time stamps allow the measurement to be combined easily and phase angle measurements to be made using a common reference.

PMUs were developed in the early 1980s [ 1 ] and are the most widely used form of synchronized measurement technology. PMUs measure voltage and current phasors at a rate of once per cycle and the IEEE C37.118 standard describes a required level of measurement performance [ 15 ] and a communication protocol [ 16 ] for these measurements. It is worth noting that this standard provides the option to include analogue and digital values into the measurement streams. This allows binary status signals and waveform measurements to be streamed using the protocol.

The architecture of a WAMS can be highly complex and [ 17 , 18 ] provides several examples of how to design a WAMS. The latency, jitter and reliability of the communication network in a WAMS is a vital aspect of ensuring that the WAMS is suitable for supporting protection functions. The communication network must be able to ensure that the measurements supplied by the WAMS to the protection functions are received not only quickly but arrive reliably and with consistent delays to ensure that the quality of the protection is sufficient.

3 Challenges faced by power system protection

3.1 overview of power system protection.

The role of power system protection is to disconnect faulty/overloaded elements to save the element from damage, prevent the fault from degrading security and to protect the surrounding area from serious danger [ 9 ].

This equipment protection is primarily delivered through breaker operations and can be broken down into primary and backup equipment protection. Primary protection avoids damage to equipment by isolating the protected equipment from the system. It is highly selective and operates in only 3~4 cycles. The relays used to deliver primary control are usually duplicated one or more times to avoid any failure to clear the fault.

Backup protection is tasked with clearing any faults that are not cleared by the primary protection. As such, it operates more slowly than primary protection, to ensure proper coordination, and is less selective. The setting of backup protection is more challenging, as it protects a larger part of the system, so is more dependent on the operating condition of the system.

The design of protection must balance two key requirements. These are dependability and security. Dependability is defined as ensuring that the protection system operates when it should. Security is defined as ensuring the protection system does not operate when it should not. However, dependability and security are opposing goals and the protection engineer must strike a balance between them.

Any protection operation can be defined according to how correct and appropriate it is. A correct relay operation is one where the relay operates as designed. An appropriate action is one that contributes positively to protecting the security of the power system. From these definitions, any relay operation can be defined according to its correctness and appropriateness [ 19 ].

In addition to equipment protection, protection is required that is tasked with preventing the partial or total loss of supply/integrity due to phenomena such as: transient angle instability, small signal instability, frequency instability, voltage instability (short and long term) and cascading outages. This system protection requires actions that go beyond breaker operations and includes actions like under frequency load shedding (UFLS). Like backup protection, system protection operates more slowly than primary protection and its settings are highly dependent on the operating conditions.

Existing protection schemes are self-contained entities that use independent local measurement chains to deliver their functionality. However, the increasing complexity of power systems has given rise to System Integrity Protection Schemes (SIPS), which use wide area measurements to deliver more complex functionality.

The measurements used by each of protection systems will vary significantly in terms of the type of measurement, the acceptable delay, the required reporting rate, the required resolution and the required accuracy.

SIPS are designed to protect the system from this specific set of contingencies [ 20 ] using a set of pre-determined actions that are designed based on offline system studies. These actions will be executed when a specific set of input conditions are satisfied [ 20 ]. For a scheme to be classed as a SIPS the actions implemented must go beyond simply isolating the faulted elements.

The conditions required to trigger a SIPS and cause it to operate can include events (e.g. the loss of a line), the system response (e.g. the measured frequency being below a threshold), or a combination thereof. Furthermore, most SIPS are armed by one condition and then triggered by another condition. The use of SIPS is now a worldwide practice [ 21 ] and an ever increasing number of these schemes are being designed and implemented.

The compatibility and coordination of protection in neighboring systems is essential, especially as it becomes more complex, far reaching and adaptive. This serves to prevent undesirable interactions [ 22 ] that may create hidden failure modes or even directly cause maloperation.

3.2 Cascade failures

Cascade failures can be described as a sequence of failures in the power system that occur one after another and each failure occurs because of the consequences of the previous failures, e.g. a sequence of line trips due to violation of thermal limits. During post-mortem analysis the initiating event of a cascade can usually be identified with ease; however, it is important to bear in mind that during operation it is harder to clearly recognize an event that will eventually initiate a cascade.

Cascade failures can occur very quickly after the initiating event and have contributed to several recent blackouts [ 7 , 8 ] and the fast, adaptive actions required for the prevention of these cascades are beyond the scope of most of the existing power system protection [ 23 ].

Local protection uses only local information and cannot consider the whole system, either its state or its needs. Therefore, it is attractive to explore the opportunity to use wide area information and real time measurements to create protection actions that are designed to protect power system security from wide area disturbances. This protection must identify the stressed conditions that may leave the system vulnerable to a cascade and the possible initiating events that exist within the system.

For example, a thermal overload can be relieved by local protection and through this the asset is protected. However, this local protection cannot assess the severity of the overload relative to the importance of the asset to system security. Removing this asset immediately may initiate a cascade of thermal overloads. In contrast, by using wide area measurements to develop an accurate view of the system state and the evolving threat to security, a wide area protection scheme could identify the importance of the asset to system security and exploit short term thermal ratings (possibly complemented with dynamic thermal line ratings [ 24 ]) to delay the local protection action and provide more time to relieve the overload by alternative means and preserve system security. Thus, wide area protection can be used to realize protection actions that adapt to the system’s needs, in terms of security, and protect against wide area disturbances and cascading failures.

Finally, the complexity of the mechanisms behind wide area disturbances and the short time frame over which they can cause system collapse may mean that their proper management is beyond a human operator, however skilled they may be [ 10 ]. In this context, automatic actions will be needed to preserve system security and wide area protection offers the opportunity to deliver these actions.

3.3 Correct but inappropriate operation of relays

The incorrect operation of protection relays has contributed to a number of cascades failures and blackouts [ 7 , 8 ]. Existing protection relays primarily use fixed characteristics that do not adapt to the true system conditions. This means that it is possible for this protection to operate correctly but inappropriately.

This problem has been exacerbated by changes in the operating practices of power systems, e.g. a greater emphasis on commercial and environmental factors. These changes have led to an increasing variety of generation mixes and load flow patterns. Therefore, the fault level and load flow pattern of the system can change quickly and the range of possible operating conditions is becoming increasingly broad. This has made the proper setting of protection far more challenging, as it is harder to determine the settings that will be applicable for all of the likely operating conditions and contingencies. This has contributed to the correct but inappropriate operation of protection relays; particularly backup protection relays [ 9 ].

3.4 Hidden failures

Despite the challenges faced by modern power system protection and the increasing complexity of protection, modern protection performs very well and almost all relay operations are correct and appropriate [ 22 ]. However, incorrect protection actions have played a role in the initiation and propagation of several major blackouts [ 7 ],[ 8 ]. A common theme in these events is the presence of hidden failures that caused a relay to operate incorrectly immediately after another protection action had been taken in their local area. A hidden failure is defined as a permanent, undetected defect in a protection relay that causes a relay to operate incorrectly and remove elements of the system as a consequence of another switching event in the system [ 25 ]. Hidden failures are random events that are not indicative of bad relay design. They do not immediately lead to an incorrect operation but will cause one when another event occurs in their local area.

Hidden failures only include those failures that cause a relay to operate incorrectly. Failures that cause the relay to not operate are not hidden failures, as they should be accommodated by redundant protection. Equally, failures that cannot be monitored are not hidden failures, they are faulty design, and temporary failures that occur, e.g. during switching, are not hidden failures.

Figure  1 presents a comparison of a hidden failure and a non-hidden failure for a three zone step distance relay that was presented in [ 26 ]. A failure of the contacts of T 3 that causes them to be permanently closed will create a hidden failure. This is because the failure of T 3 does not cause an immediate maloperation, as Z 3 must also be closed. However, in the event of a fault the line will be immediately tripped without delay when Z 3 closes in the presence of any fault in Zones 1-3.

Example of a hidden failure (HF) and a non-hidden failure (NHF) for a three zone step distance relay-repeated from [ 26 ]

In contrast, a failure of the contacts of Z 1 that causes them to be permanently closed will not create a hidden failure. This is because at the instant of the failure the line will be tripped. Whilst this is a maloperation, it is not a hidden failure, as immediately caused the line trip.

Possible hidden failures include: relay contacts that are always open or closed, timers that operate instantaneously regardless of the set delay, outdated settings, settings that are unsuitable for the prevailing conditions, and human error in relay coordination [ 26 ].

Hidden failures are a particular threat because they require another event in the local area to reveal them. This means that a hidden failure and its triggering event represent two related failures, which is a far more severe threat than two random, unrelated failures. Furthermore, the triggering event itself is usually a sign that the power system is experiencing stressed conditions. These factors mean that hidden failures inherently threaten to contribute to a cascade of failures in their local area.

This local area was more strictly defined as a region of vulnerability in [ 25 ] and will vary significantly for different modes of hidden failure in different elements.

The design of any protection scheme will directly influence the likelihood of it experiencing hidden failures [ 26 ]. The nature of wide area protection schemes may mean that their region of vulnerability could be significantly larger than those seen for existing protection. As such, the hidden failure modes and region of vulnerability of a wide area protection scheme should be rigorously assessed to ensure that their presence does not weaken the protection of the system as a whole [ 27 ].

The greater complexity of SIPS and WAP, compared to traditional protection, will mean that the task of analyzing them for hidden failures will be more challenging. A particular challenge involved in analyzing WAP will be the analysis of the wide area monitoring and communication networks on which they depend. These networks can be highly complex and depend on a wide variety of multi-vendor hardware and technologies. Furthermore, the broader scope of actions available to a SIPS and WAP (e.g. system separation) will mean that the impact of any hidden failure modes may be far greater than it would be for other protection elements.

Bearing in mind the increased complexity of analyzing SIPS and WAP to identify hidden failures and the greater consequences of their maloperation; it is particularly important that they are designed with the minimization of hidden failure modes in mind alongside the ability to self-diagnose failures and adapt to them. These considerations should extend beyond the original design to include the development of maintenance procedures.

Hidden failures can only be detected when they cause an incorrect operation or when the faulty element is tested. Ongoing maintenance, calibration and review of protection could identify existing hidden failures and correct them [ 19 ] and recent work has presented a number of such methods [ 28 ]. However, given the number of protection elements, this ongoing task may be difficult to deliver with the resources available. Therefore, it may be attractive to develop more methods for exploiting the ability of digital relays to self-diagnose the presence of failure modes. Furthermore, WAMS based concepts for detecting these failures, like those proposed in [ 29 ] that can identify such failures may be necessary.

However, it is known that maintenance is a source of hidden failures. Therefore, it is important to develop WAP concepts that can help to limit the impact of hidden failures when they are revealed. Furthermore, recent work, e.g. [ 30 , 31 ] has incorporated hidden failures into the statistical modelling of power system reliability using expert systems, importance sampling, neural networks and fuzzy logic. A review of this work is provided in [ 29 ].

4 Enhancing protection with wide area monitoring

The overall objective of using wide area monitoring to enhance protection is to create new protection concepts that will make blackouts less likely to occur and less intense when they do occur. The key areas in which WAM can contribute to power system protection are as follows.

Avoiding inappropriate relay settings for the prevailing system conditions

Managing wide area disturbances

Mitigating the impact of hidden failures

Ensuring a suitable balance between the security and dependability of protection

The goal of protection is to protect individual elements of the power system from damage and to protect the security of the power system itself.

In the case of primary equipment protection there is very little role for the use of wide area monitoring. This is because primary protection must reliably deliver a very fast response for any fault on the element that it protects. However, the slower speed of response required for backup protection and the fact that it protects a zone of the system means that wide area monitoring can be a useful tool for improving its performance.

The most effective means for ensuring that the system will survive extreme conditions and wide area disturbances is a high degree of built in redundancy and strength [ 32 ]. However, this over engineering of the system is not compatible with the economic and environmental demands placed upon modern power systems. Therefore, a significant role for wide area monitoring enhanced protection may be to enable system operators to deliver the existing level of security and reliability in these new operating conditions.

Wide area measurements offer the potential to create supervisory schemes for backup protection, more advanced forms of system protection and entirely new protection concepts. Examples of these protection functions include [ 32 ]:

Adaptive relays that update their settings as the system state changes

Improved protection of multi terminal lines

Adaptive end of line protection that monitors the remote breaker, if it is open the under reaching Zone 1 is replaced with an instantaneous characteristic

Temporarily adapt relay settings to prevent maloperation during cold load pickup

Use the ability of digital relays to self-monitor to identify hidden failures and use the hot swap functionality offered by IEC 61850 to remove them

Intelligent controlled islanding that preempts an uncontrolled system separation by implementing an adaptive controlled separation

The remainder of this section discusses some of the opportunities for wide area monitoring enhanced protection in more detail.

4.1 Alarming against the risk of relay characteristic penetration

The objective of this application is to detect when the impedance observed by a relay approaches the relay characteristic under non-faulty conditions. This information is then used to alarm protection engineers to a relay setting that is potentially unsuitable [ 32 ].

This concept does not directly improve the performance of protection or use wide area measurements. However, it does use the communication network that is necessary for wide area monitoring to generate valuable information that will help protection engineers to improve the security and reliability of protection. This method could be applied to critical relays that are vulnerable to load encroachment and/or power swings or to relays that will have more severe consequences in the event of any maloperation.

4.2 Preventing load encroachment

The loadability of an impedance relay is the maximum load that can be distinguished from a fault. This is highly dependent on voltage at the bus and reactive power flows, which can vary dramatically during stressed conditions and power swings. Heavily loaded lines may encroach on the settings of relays and cause an incorrect and inappropriate tripping operation. This load encroachment of impedance relays played a role in recent blackouts [ 7 , 8 ] and arises because the relay setting is a compromise between the desired setting level and the maximum anticipated load at the relay locations. This compromise must accommodate a wide range of possible system conditions, loadings and contingencies.

This compromise is vulnerable to unforeseen conditions, as it is based on offline simulations of the credible operating conditions and contingencies. As such, the relay setting would only be suitable provided that the assumptions made when it was set hold true. With the more variable nature of modern power systems and the introduction of significant intermittent generation, it is likely that this compromise would become ever more inefficient, as the variation between the maximum loading and the normal loading would become more significant and variable [ 33 ]. With the computational power of digital relays this can be overcome by using real time measurements of the load to prevent load encroachment by compensating the relay input for the load current [ 32 ].

4.3 Adjusting the balance between the security and dependability of protection

Balancing the demands of dependability and security is one of the greatest challenges during the design of protection. Existing protection is designed to favor dependability [ 34 ]. This preference for dependability is attractive during healthy operation when the threat of an uncleared fault is severe and the system can easily survive the loss of a single element, due to the inherently high level of redundancy in a healthy power system.

However, during a wide area disturbance, this preference for dependability can result in incorrect and inappropriate tripping operations. This is a major threat to a stressed system, as the loss of a single element can accelerate the systems descent into a cascade failure and even blackout.

Therefore, it would attractive to shift the balance of this compromise toward security during stressed conditions, i.e. when the conditions encountered (e.g. power swings) can increase the likelihood of maloperations and reveal hidden failures. The highly redundant nature of power system protection means that there are many different possible ways of combining the outputs of the various relays to select the balance between dependency and security.

Wide area measurements could be used to detect that the system has entered a stressed condition and then adjust the protection philosophy to shift the balance away from dependability and toward security. In Fig.  2 , this is achieved by swapping between an OR operation, majority voting and an AND operation. The supervisory signal selects the logical combination used to determine the breaker trip signal from each individual relays trip signals. Adapted from [ 35 ].

The use of WAM to vary the balance between dependability and security

This approach would slightly increase the likelihood of a fault not being cleared. However, with the existing protection approach, the probability of a fault not being cleared is very low. Therefore, this small increase in the probability of not clearing a fault is acceptable, as it offers a significant reduction in the likelihood of inappropriate protection action from exacerbating stressed conditions and driving the system closer to a blackout [ 22 ].

This form of adaptive protection based on wide area measurements could be an effective solution to the challenge posed by hidden failures. By requiring multiple relays to approve any tripping, it would prevent a single hidden failure in any one of these relays from causing an incorrect and inappropriate tripping operation. However, as hidden failures can appear in any element of a protection scheme, any increase in the complexity of protection must be thoroughly assessed in terms of their own modes of failure, both hidden and non-hidden.

4.4 Supervision of back-up zones

The maloperation of zone 3 relays was identified as a significant contributing factor to recent blackouts [ 7 ][ 36 ]. The unusual load currents and power swings observed during wide area disturbances can cause these relays to operate undesirably. Examples of the system behavior that can cause maloperation of a relay are shown in Fig.  3 .

Examples of dynamic conditions that can cause maloperation of distance relays [ 32 ]

This vulnerability has led to some calls for zone 3 to be abandoned; but most authors agree that this is too extreme and instead wide area measurements should be used to improve the performance of backup protection [ 10 ].

An example of how this can be achieved is the supervision of backup protection using pick up signals from remote PMUs [ 35 ]. An example of this is depicted in Fig.  4 . Furthermore, measurements of negative sequence currents may be used to further improve this concept.

Supervision of backup relay operation using remote PMUs to check for a fault in Zone 3 [ 35 ]

The remote PMUs are installed within the protection zone of the backup relay and monitor the current at these remote locations. These devices implement a simple pick up characteristic and communicate a binary pick up signal to the backup relay. If the backup relay characteristic is violated but none of the remote devices have picked up, then it can be concluded that no fault has occurred and the backup relay operation can be blocked. This prevents load swings during extreme conditions from being misinterpreted as faults and helps prevent the maloperation of backup relays from allowing a wide area disturbance to spread through the system.

The enhancement of backup protection has been a particular focus of recent work and methods based on wide area impedances and current indices [ 37 ], net current injection into predefined zones [ 38 ], and voltage measurements [ 39 ], have been proposed. Furthermore, recent work [ 40 ] has presented a scheme that is designed for the specific and challenging case of series compensated lines. These methods can either supervise or substitute existing zone 3 relays, although further work is required in the area of communication redundancy [ 39 ]. The majority of these recent methods are WAP based; however, some are not and [ 41 ] use an energy function derived from three phase measurements and the local phase angle to block zone 3 operation.

4.5 Intelligent under frequency load shedding

Load shedding is the traditional last line of defense against extreme under frequency conditions. Current practice is mostly for this shedding to be delivered using a sequence of stages of shedding that are triggered when a certain frequency threshold is violated [ 42 ]. Shedding load more quickly after a loss of infeed is recognized as an effective means for limiting the frequency deviation with a reduced amount of load shedding [ 43 ]. However, balancing the benefits of an increased speed of response against the risk of unnecessary shedding is a challenge.

In isolated power systems frequency control is becoming an increasing area of concern. The displacement of traditional synchronous generation with asynchronous generation is reducing system inertia and allowing larger, faster frequency deviations to occur [ 44 , 45 ].

Extensive research has been undertaken to create more advanced load shedding schemes that use wide area measurements to reduce the amount of load shed by:

Adapting the amount of load shed to the prevailing system conditions, e.g. inertia

Initiating the load shedding more quickly

Initiating the load shedding more quickly can be achieved by using event based signals (e.g. the loss of a major interconnector or generator) or by using more complex triggering signals (e.g. triggering based on rate of change of frequency). Furthermore, the amount of load shed can be adapted to the size of the disturbance and system inertia using wide area measurements.

Examples of this work include the adaption of shedding based on measurements of rate of change of frequency (RoCoF) immediately after the disturbance [ 46 ] and [ 47 ]. However, accurately measuring the RoCoF quickly is a challenge and [ 48 ] identifies a number of potential threats to its successful use in adaptive load shedding. Other work addresses load shedding as an optimisation problem that can be solved using genetic algorithms [ 49 ] and neural networks [ 50 ]. Recent work has incorporated aspects of dynamic security assessment and prediction of the frequency response [ 51 ]. Furthermore, some authors have attempted to reflect the impact of UFLS on the system as a whole, e.g. the changes in voltage, reactive flows [ 52 ] and line loading [ 53 ].

4.6 Adaptive out-of-step relaying

Out of step conditions and system separation are key precursors to system collapse and blackouts. As the formation of an electrical center approaches the system will experience extreme power swings that will further exacerbate stressed conditions and drive the system closer to collapse. Therefore, it is imperative that any potential out of step condition is quickly recognized and prevented; this is the role of out of step relays.

Predicting out of step conditions with local measurements is a challenging task that depends upon settings that are selected using transient simulation of various contingencies and system conditions [ 35 ].

Based on these simulations two zones are defined for impedance relays that are installed close to the anticipated electrical center and any violation of the inner zone denotes an out of step condition [ 35 ].

However, this is only a reliable approach for simple systems that can be characterized as two areas that are swinging against one another, e.g. the system in operation for the Florida—Georgia Interconnection [ 54 ].

In more complex systems the power flows and synchronizing coefficients vary too much for the assumed characteristics to remain accurate for long. Therefore, the relay characteristic will become either too sensitive, allowing inappropriate operation, or insensitive, preventing the relay from ever operating. Although the relay setting could be updated as conditions vary, ongoing adjustment of protection in this way is undesirable; as it will likely serve as a source of hidden failures (as any maintenance of protection schemes can be).

A wide area protection scheme could be developed that monitors the positive sequence voltages across the system. These synchronized real time measurements can be used to predict if regions of the system are approaching an out of step condition [ 35 ]. This prediction could be used to initiate a controlled separation of the areas that are losing synchronism [ 10 ] or, if the prediction is available sufficiently in advance, actions could be taken to prevent the out of step condition from occurring and avoid system separation entirely. The challenge faced when developing such a scheme would be selecting the measurement locations and developing the algorithms for achieving robust real time coherency determination when the coherent generator groups are variable.

4.7 System integrity protection schemes (SIPSs)

SIPS protect power system security from extreme contingencies or wide area disturbances that are beyond the scope of traditional protection. The increasing availability and maturity of real time wide area measurements has enabled the creation of more advanced SIPS that are able to protect power systems from wide area disturbances for a wide range of operating conditions.

The stages involved in the execution of a SIPS are: ① Identification and prediction of stressed conditions, ②Classification of the threat to system security, ③ Decisions and actions, ④ Coordination, and ⑤ Correction.

Examples of SIPS include [ 55 ]: generator rejection, load rejection, under frequency and voltage load shedding, system separation, dynamic braking, and turbine valve control. In [ 10 ] several operational SIPS are described.

The actions available to a SIPS include [ 56 ]: load shedding, generation start up/rejection, switching of shunt reactors, line tripping, tap changes, adjusting controller set points, tap blocking, controlled islanding, HVDC control and switching of braking resistors.

SIPSs, like all protection, take corrective actions in an attempt to protect the power system from the consequences of contingencies. However, the increasing attraction toward SIPS is because of their ability, through the availability of real time wide area measurements, to identify complex emerging threats to the power system and respond to them quickly and decisively in a way that protection other cannot. For example, event based SIPS can respond immediately after a severe contingency, or combination of contingencies, rather than waiting for the inevitable degradation of the system state. In contrast, response based SIPS can use real time measurements of the system state after a contingency to assess the need for a response and adapt the nature of any response to the true system state. Furthermore, event based and response based decision making can be combined to create complex SIPS that can deliver fast and adaptive protection actions for a wide range of system conditions and contingencies.

However, the severity of the contingencies that SIPS are designed to protect against and the highly intrusive nature of many of the actions available to them mean that SIPS face onerous requirements in terms of both dependability and security [ 56 ]. For example, a failure to operate could result in a wide area disturbance going unchecked, most probably leading to a blackout, and operating unnecessarily could cause a blackout when the system was operating in a healthy condition.

The complexity of novel SIPS and their proliferation makes the proper coordination of the various SIPS in a power system a significant task. This is vital because the maloperation of a SIPS could have far reaching consequences. Furthermore, the wide area nature of certain SIPS will mean that the SIPS of neighboring systems must also be coordinated.

4.8 Application of WAP to distribution networks

The changing nature of power systems and the possible benefits of wide area protection also extend to the protection of the distribution system. The changes faced by distribution networks include the connection of energy storage, electric vehicles, smart meters, demand side participation and the connection of distributed generation (DG). Furthermore, these changes must be faced with an ageing asset base and an increasing total load.

The increasing connection of DG is a particularly significant change, as it has resulted in distribution networks undergoing a radical change from single source, radial systems to more complex multi-source systems. This has introduced a number of threats to distribution system protection including reverse power flows and the contribution of DG to fault currents. The nature of the threat varies with the relative position of the fault, the relay and the DG, but can include false tripping, and a loss of sensitivity or selectivity [ 57 , 58 ]. Also, high fault levels at the distribution level could allow fault currents to exceed those that can be safely interrupted by the available protection.

These threats have meant that IEEE std 1547 recommends the disconnection of DG during faults. This is an obvious and significant barrier to DG playing a significant role in system operation under stressed conditions. To overcome this barrier new protection concepts are required that offer superior performance. Wide area protection that uses information from multiple locations to quickly and selectively clear the fault in these more complex distribution networks is an attractive solution. The new concepts proposed include:

The introduction of directional overcurrent relays to replace the overcurrent relays that are prevalent in existing systems [ 59 ];

The use of multi agent systems that can monitor multiple locations and make adaptive relaying decisions [ 60 ]

Enhanced pilot protection [ 61 ]

Enhanced converter response during faults [ 62 ]

Thermal protection relays that use an inference engine to combine dynamic ratings and coordination of DG to manage loading [ 63 ]; and

The use of negative sequence current dot protection (I 2 DP).

WAP at the distribution level will depend upon similar infrastructure and technology as those systems at the transmission level. However, the smaller angular separation across a distribution network means that measurement of angles on the distribution network is more demanding than it is at the transmission level. A particularly important enabler for these new protection principles are micro-processor relays that can vary their settings easily and the IEC 61850 standard will be essential for fully realizing the capabilities of these devices and delivering the protection needs of future distribution systems [ 62 ].

Another motivation for WAP at the distribution level is its role as an enabler for adaptive control, e.g. automatic network reconfiguration that reduces the frequency and length of customer interruptions, manages circuit loading, and limits the fault level [ 57 ]. Adaptive control of the distribution network is becoming increasingly necessary to reduce barriers to DG, make best use of the installed DG and through this help to deliver a low carbon future. This adaptive control and other measures form part of a move toward the creation of active distribution networks [ 64 ] and existing protection is not compatible with many of these adaptive control measures [ 57 ].

Finally, the desire to deliver ever improving quality and security of supply to customers has led to increasing pressure for the design of protection to ensure that any interruption of supply is minimized [ 65 ].

The creation of ad-hoc or planned microgrids is an effective means for maintaining supply or more quickly restoring supply after faults in the distribution system [ 58 ]. However, the challenges faced by distribution networks are equally, if not more so, relevant for microgrids [ 58 ]. A particular challenge is that the protection of microgrids must function correctly for both an autonomous microgrid and a non-autonomous microgrid, which will require a significant degree of adaptation and reconfiguration.

5 Conclusion

WAM offers a wide variety of opportunities for enhancing the backup protection and system protection of modern power systems. These enhancements can contribute to reducing the likelihood of the maloperation of backup relays, limiting the impact of hidden failures and creating new tools for managing wide area disturbances. These benefits indicate that the main role of wide area monitoring as part of protection is improving the resilience of power systems against stressed conditions and wide area disturbances, not the isolation of individual faults. The well-considered deployment of these new concepts should reduce the frequency and intensity of blackouts and enable more rapid service restoration.

The increasing vulnerability of power systems to wide area disturbances and the short time over which these extreme events can cause system collapse may mean that automatic, adaptive actions, like those offered by system integrity protection schemes, may be the only effective means to protect power system security in the future.

However, if these new concepts are to be deployed then significant efforts must be undertaken to understand their potential for hidden failures and unwanted interactions. A particular focus should be on how to coordinate these more complex protections schemes with one another; both within a system and between neighboring systems.

Finally, the performance of the supporting communication infrastructure, in terms of latency, jitter, redundancy and cyber security, will determine the performance of any form of wide area monitoring based protection. As such, the architecture used for delivering this enhanced protection will be an important factor in determining its success.

Phadke AG, Thorp JS, Adamiak MG (1983) A new measurement technique for tracking voltage phasors, local system frequency, and rate of change of frequency. IEEE Trans Power Appar Syst 102(5):1025–1038

Article   Google Scholar  

Terzija V, Valerde G, Cai DY et al (2011) Wide-area monitoring, protection, and control of future electric power networks. Proc IEEE 99(1):80–93

Chakrabarti S, Kyriakides E, Bi TS et al (2009) Measurements get together. IEEE Power Energy Mag 7(1):41–49

Phadke AG (2002) Synchronized phasor measurements-a historical overview. In: Proceedings of the IEEE PES transmission and distribution conference and exhibition 2002: Asia Pacific, vol 1, Yokohama, Japan, 6–10 Oct 2002, pp 476–479

Phadke AG (2008) The wide world of wide-area measurement. IEEE Power Energy Mag 6(5):52–65

Phadke AG, Thorp JS (2008) Synchronized phasor measurements and their applications. Springer Science+Business Media, Boston

Book   MATH   Google Scholar  

US-Canada Power System Outage Task Force (2004) Blackout 2003: Final report on the August 14, 2003 blackout in the United States and Canada: Causes and recommendations. Office of Electricity Delivery & Energy Reliability, Washington, DC

Google Scholar  

Report of the Enquiry Committee on grid disturbance in northern region on 30th July 2012 and in northern, eastern & north-eastern region on 31st July 2012. The Enquiry Committee, Ministry of Commerce and Industry, Government of India, New Delhi, India, 2012

Karlsson D, Broski L, Ganesan S (2004) Maximizing power system stability through wide area protection. In: Proceedings of the 57th annual conference for protective relay engineers, College Station, TX, USA, 30 Mar–1 Apr 2004, pp 403–418

Horowitz SH, Phadke AG (2006) Third zone revisited. IEEE Trans Power Deliv 21(1):23–29

Ding L, Gonzalez-Longatt FM, Wall P et al (2013) Two-step spectral clustering controlled islanding algorithm. IEEE Trans Power Syst 28(1):75–84

Phadke AG, Thorp JS (2010) Communication needs for wide area measurement applications. In: Proceedings of the 5th international conference on critical infrastructure (CRIS’10), Beijing, China, 20–22 Sept 2010, 7 pp

Begovic M (2004) Wide area protection and emergency control. In: Proceedings of the 2004 IEEE PES power systems conference and exposition, vol 3, New York, NY, USA, 10–13 Oct 2004, pp 1776–1777

Shi BN, Zhang DN, Hu J (2015) Preliminary investigation in wide area protection implementation using IEEE 1588 precision time protocol. In: Proceedings of the IEEE international symposium on precision clock synchronization for measurement, control, and communication (ISPCS’15), Beijing, China, 11–16 Oct 2015, pp 43–47

IIEEE Standard C37.118.1 (2011) IEEE standard for synchrophasor measurements for power systems

IEEE Standard C37.118.2 (2001) IEEE standard for synchrophasor measurements for power systems

Terzija V, Regulski P, Kunjumuhammed LP, et al (2011) FlexNet wide area monitoring system. In: Proceedings of the 2011 IEEE power and energy society general meeting, San Diego, CA, USA 24–29 Jul 2011, 7 pp

Cai DY, Wall P, Osborne M et al (2016) Roadmap for the deployment of WAMPAC in the future GB power system. IET Gener Transm Distrib 10(7):1553–1562

Horowitz SH, Phadke AG (2006) Blackouts and relaying considerations—relaying philosophies and the future of relay systems. IEEE Power Energy Mag 4(5):60–67

System protection schemes in power networks. Task Force 38.02.19, CIGRE, Paris, France, 2001

Defense plan against extreme contingencies. Task Force C2.02.24, CIGRE, Paris, France, 2007

Thorp JS, Phadke AG (1999) Protecting power systems in the post-restructuring era. IEEE Comput Appl Power 12(1):33–37

Begovic M, Novosel D, Karlsson D et al (2005) Wide-area protection and emergency control. Proc IEEE 93(5):876–891

Cong YH, Wall P, Regulski P et al (2015) On the use of dynamic thermal line ratings for improving operational tripping schemes. IEEE Trans Power Deliv (to be published)

Tamronglak S, Horowitz SH, Phadke AG et al (1996) Anatomy of power system blackouts: Preventive relaying strategies. IEEE Trans Power Deliv 11(2):708–715

Elizondo DC, De La Ree J, Phadke AG, et al (2001) Hidden failures in protection systems and their impact on wide-area disturbances. In: Proceedings of the 2001 IEEE power engineering society winter meeting, vol 2, Columbus, OH, USA, 28 Jan–1 Feb 2001, pp 710–714

Phadke AG, Thorp JS (1996) Expose hidden failures to prevent cascading outages [in power systems]. IEEE Comput Appl Power 9(3):20–23

Zheng TY, Hu ED, Yang H, et al (2016) Fast, in situ demagnetization method for protection current transformers. IEEE Trans Magn (to be published)

Kundu P, Pradhan AK (2015) Online identification of protection element failure using wide area measurements. IET Gener Transm Distrib 9(2):115–123

Lei HT, Singh C (2015) Incorporating protection systems into composite power system reliability assessment. In: Proceedings of the 2015 IEEE Power and Energy Society general meeting, Denver, CO, USA, 26–30 Jul 2015, 5 pp

Liu XD, Shahidehpour M, Cao YJ et al (2015) Risk assessment in extreme events considering the reliability of protection systems. IEEE Trans Smart Grid 6(2):1073–1081

Horowitz SH, Phadke AG (2003) Boosting immunity to blackouts. IEEE Power Energy Mag 1(5):47–53

Horowitz SH, Phadke AG, Renz BA (2010) The future of power transmission. IEEE Power Energy Mag 8(2):34–40

Horowitz SH, Phadke AG, Thorpe JS (1988) Adaptive transmission system relaying. IEEE Trans Power Deliv 3(4):1436–1445

De La Ree J, Centeno V, Thorp JS et al (2010) Synchronized phasor measurement applications in power systems. IEEE Trans Smart Grid 1(1):20–27

Horowitz SH, Novosel D, Madani V et al (2008) System-wide protection. IEEE Power Energy Mag 6(5):34–42

Kundu P, Pradhan AK (2014) Synchrophasor-assisted zone 3 operation. IEEE Trans Power Deliv 29(2):660–667

Kalantar Neyestanaki M, Ranjbar AM (2015) An adaptive PMU-based wide area backup protection scheme for power transmission lines. IEEE Trans Smart Grid 6(3):1550–1559

Zare J, Aminifar F, Sanaye-Pasand M (2015) Synchrophasor-based wide-area backup protection scheme with data requirement analysis. IEEE Trans Power Deliv 30(3):1410–1419

Jena MK, Samantaray SR, Panigrahi BK (2015) A new wide-area backup protection scheme for series-compensated transmission system. IEEE Syst J (to be published)

Nayak PK, Pradhan AK, Bajpai P (2015) Secured zone 3 protection during stressed condition. IEEE Trans Power Deliv 30(1):89–96

IEEE Standard PC37.117/D7.0 (2006) Guide for the application of protective relays used for abnormal frequency load shedding and restoration sponsored by the Power Systems Relaying Committee of the IEEE Power Engineering Society

Lokay HE, Burtnyk V (1968) Application of underfrequency relays for automatic load shedding. IEEE Trans Power Appar Syst 87(3):776–783

Electricity ten year statement 2015. National Grid, London, UK, 2015

Wall P, Terzija V (2014) Simultaneous estimation of the time of disturbance and inertia in power systems. IEEE Trans Power Deliv 29(4):2018–2031

Terzija V (2007) Adaptive underfrequency load shedding based on the magnitude of the disturbance estimation. In: Proceedings of the IEEE Power Engineering Society general meeting, Tampa, FL, USA, 24–28 Jun 2007, 1 pp

Rudez U, Mihalic R (2011) Monitoring the first frequency derivative to improve adaptive underfrequency load-shedding schemes. IEEE Trans Power Syst 26(2):839–846

Rudez U, Mihalic R (2011) Analysis of underfrequency load shedding using a frequency gradient. IEEE Trans Power Deliv 26(2):565–575

Hong YY, Wei SF (2010) Multiobjective underfrequency load shedding in an autonomous system using hierarchical genetic algorithms. IEEE Trans Power Deliv 25(3):1355–1362

Hsu CT, Kang MS, Chen CS (2005) Design of adaptive load shedding by artificial neural networks. IEE Proc Gener Transm Distrib 152(3):415–421

Rudez U, Mihalic R (2015) WAMS based underfrequency load shedding with short-term frequency prediction. IEEE Trans Power Deliv (to be published)

Tang JJ, Liu JQ, Ponci F et al (2013) Adaptive load shedding based on combined frequency and voltage stability assessment using synchrophasor measurements. IEEE Trans Power Syst 28(2):2035–2047

Novosel D, King RL (1994) Using artificial neural networks for load shedding to alleviate overloaded lines. IEEE Trans Power Deliv 9(1):425–433

Centeno V, Vhadke AG, Edris A (1997). Adaptive out-of-step relay with phasor measurement. In: Proceedings of the 6th international conference on developments in power system protection, Nottingham, UK, 25–27 Mar 1997, pp 210–213

Anderson PM, Le Reverend BK (1996) Industry experience withspecial protection schemes. IEEE Trans Power Syst 11(3):1166–1179

Crossley P, Ilar F, Karlsson D (2001) System protection schemes in power networks: Existing installations and ideas for future development. In: Proceedings of the 7th international conference on developments in power system protection, Amsterdam, Netherland, 9–12 Apr 2001, pp 450–453

Allen W (2009) Effects of wide-area control on the protection and operation of distribution networks. In: Proceedings of the 2009 power systems conference (PSC’09), Clemson, SC, USA, 10–13 Mar 2009, 10 pp

Conti S (2011) Protection issues and state of the art for microgrids with inverter-interfaced distributed generators. In: Proceedings of the 2011 international conference on clean electrical power (ICCEP’11), Ischia, Italy, 14–16 Jun 2011, pp 643–647

Choudhury S, Mohapatra L, Rout PK (2015) A comprehensive review on modeling, control, protection and future prospects of microgrid. In: Proceedings of the 2015 international conference on electrical, electronics, signals, communication and optimization (EESCO’15), Visakhapatnam, India, 24–25 Jan 2015, 6 pp

Qin LJ, Wang Y, Hao CJ, et al (2011) Multi-agent system wide area protection considering distributed generation impact. In: Proceedings of the 2011 international conference on advanced power system automation and protection (APAP’11), vol 1, Beijing, China, 16–20 Oct 2011, pp 549–553

Thirumalai A, Liu X, Karady GG (2011) Ultra fast pilot protection of a looped distribution system. In: Proceedings of the 2011 IEEE Trondheim PowerTech conference, Trondheim, Norway, 19–23 Jun 2011, 8 pp

Laaksonen HJ (2010) Protection principles for future microgrids. IEEE Trans Power Electron 25(12):2910–2918

Yip HT, An C, Lloyd GJ, et al (2010) Dynamic thermal rating and active control for improved distribution network utilisation. In: Proceedings of the 10th IET international conference on developments in power system protection (DPSP’10), Manchester, UK, 29 Mar–1 Apr 2010, 5 pp

Zhang F, Mu LH (2015) Wide-area protection scheme for active distribution networks based on phase comparison. In: Proceedings of the 5th international conference on electric utility deregulation and restructuring and power technologies (DRPT’15), Changsha, China, 26–29 Nov 2015, pp 927–932

Wen A, Zhao MY, Huang WF, et al (2015) Design and development of wide area protection and emergency control for application in distribution networks of embedded generation. In: Proceedings of the power and energy society general meeting, Denver, CO, USA, 26–31 Jul 2015, 5 pp

Download references

Author information

Authors and affiliations.

Virginia Polytechnic Institute and State University, 900 N. Glebe Rd., Arlington, VA, 22203, USA

Arun G. PHADKE

Department of Electrical and Electronic Engineering, University of Manchester, Manchester, M13 9PL, UK

Peter WALL & Vladimir TERZIJA

School of Electrical Engineering, Shandong University, 17923 Jingshi Road, Jinan, 250061, Shandong Province, China

You can also search for this author in PubMed   Google Scholar

Corresponding authors

Correspondence to Peter WALL or Vladimir TERZIJA .

Additional information

CrossCheck date: 14 June 2016

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License ( http://creativecommons.org/licenses/by/4.0/ ), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Reprints and permissions

About this article

PHADKE, A.G., WALL, P., DING, L. et al. Improving the performance of power system protection using wide area monitoring systems. J. Mod. Power Syst. Clean Energy 4 , 319–331 (2016). https://doi.org/10.1007/s40565-016-0211-x

Download citation

Received : 23 March 2016

Accepted : 14 June 2016

Published : 13 July 2016

Issue Date : July 2016

DOI : https://doi.org/10.1007/s40565-016-0211-x

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Backup protection
• Hidden failures
• Power system protection
• System integrity protection schemes
• Wide area monitoring
• Wide area protection
• Find a journal
• Publish with us
• Track your research

IEEE Account

• Change Username/Password
• Update Address

Purchase Details

• Payment Options
• Order History
• View Purchased Documents

Profile Information

• Communications Preferences
• Profession and Education
• Technical Interests
• US & Canada: +1 800 678 4333
• Worldwide: +1 732 981 0060
• Contact & Support
• About IEEE Xplore
• Accessibility
• Terms of Use
• Nondiscrimination Policy
• Privacy & Opting Out of Cookies

A not-for-profit organization, IEEE is the world's largest technical professional organization dedicated to advancing technology for the benefit of humanity. © Copyright 2024 IEEE - All rights reserved. Use of this web site signifies your agreement to the terms and conditions.

Academia.edu no longer supports Internet Explorer.

To browse Academia.edu and the wider internet faster and more securely, please take a few seconds to  upgrade your browser .

•  We're Hiring!
•  Help Center

Power System Protection Engineering

• Most Cited Papers
• Most Downloaded Papers
• Newest Papers
• Last »
• Electric Power Systems Follow Following
• Electrical Engineering Follow Following
• Reactive Power Follow Following
• Transmission and Distribution Follow Following
• Electrical Substation Design Follow Following
• Power System Follow Following
• Power Quality Follow Following
• Microgrid Follow Following
• Artificial Neural Networks Follow Following
• Distance Relay Follow Following

Enter the email address you signed up with and we'll email you a reset link.

• Academia.edu Journals
•   We're Hiring!
•   Help Center
• Find new research papers in:
• Health Sciences
• Earth Sciences
• Cognitive Science
• Mathematics
• Computer Science
• Academia ©2024

• Skip to main content
• Skip to search
• Skip to footer

Products and Services

Cisco Security

Master your goals. innovate. we'll tackle threats..

Get powerful security across all your networks, cloud, endpoints, and email to protect everything that matters, from anywhere.

If it's connected, you're protected

Cisco Security “The Hacker”

More connected users and devices creates more complexity. Cisco Security Cloud makes security easier for IT and safer for everyone anywhere security meets the network.

Deliver smarter, stronger security

Protect your organization across a multicloud environment, while simplifying security operations, improving scalability, and driving data-informed outcomes, powered by Cisco Talos.

Unlock better user experiences

Create a seamless experience that frustrates attackers, not users, by granting access from any device, anywhere, and adding more proactive security controls.

Deliver cost-effective defenses

Improve ROI by consolidating vendors, reducing complexity and integrating your security.

Strengthen security resilience

Unified, end-to-end protection maximizes value, minimizes risk, and closes security gaps everywhere to defend against evolving threats. Protect access, apps, and innovation across your network to secure your future.

Cisco Secure Firewall

Better visibility and actionable insights across networks, clouds, endpoints, and email allows users to respond confidently to the most sophisticated threats at machine scale.

Featured security products

Cisco hypershield.

A new groundbreaking security architecture that makes hyperscaler technology accessible to enterprises of all sizes and delivers AI-native security for modern data centers and cloud.

Cisco Secure Access (SSE)

A converged cybersecurity solution, grounded in zero trust, that radically reduces risk and delights both end users and IT staff by safely connecting anything to anywhere.

Detect the most sophisticated threats sooner across all vectors and prioritize by impact for faster responses.

Cisco Multicloud Defense

Gain multidirectional protection across clouds to stop inbound attacks, data exfiltration, and lateral movement.

Secure applications and enable frictionless access with strong MFA and more. Establish user and device trust, gain visibility into devices, and enable secure access to all apps.

Cisco Identity Services Engine (ISE)

Simplify highly secure network access control with software-defined access and automation.

Security Suites delivered by Cisco Security Cloud

Cisco User Protection Suite

Get secure access to any application, on any device, from anywhere. Defend against threats targeting users and deliver seamless access for hybrid work.

Cisco Cloud Protection Suite

Secure your apps and data with a powerful, flexible framework for a hybrid and multicloud world.

Cisco Breach Protection Suite

Secure your business by investigating, prioritizing, and resolving incidents through unified defense and contextual insights from data-backed, AI-powered security.

Customer stories and insights

Global partnerships fight to end child exploitation together.

"Marriott has long championed human rights and human trafficking awareness. Combating CSAM is an important extension of that work. The IWF provided the level of rigor we needed in a URL list, and Cisco's security technology provided the means to easily apply it."

Abbe Horswill, Director, Human Rights and Social Impact

Company: Marriott International

The NFL relies on Cisco

"From securing stadiums, broadcasts, and fans to protecting the largest live sporting event in America, the right tools and the right team are key in making sure things run smoothly, avoiding disruptions to the game, and safeguarding the data and devices that make mission-critical gameday operations possible."

Add value to security solutions

Cisco Security Enterprise Agreement

Instant savings

Experience security software buying flexibility with one easy-to-manage agreement.

Services for security

Let the experts secure your business

Get more from your investments and enable constant vigilance to protect your organization.

Sharpen your security insights

Cisco Cybersecurity Viewpoints

Set your vision to a more secure future with Cisco Cybersecurity Viewpoints. With specialized content from podcasts to industry news, you'll walk away with a deeper understanding of the trends, research, and topics in our rapidly changing world.

Information

• Author Services

Initiatives

You are accessing a machine-readable page. In order to be human-readable, please install an RSS reader.

All articles published by MDPI are made immediately available worldwide under an open access license. No special permission is required to reuse all or part of the article published by MDPI, including figures and tables. For articles published under an open access Creative Common CC BY license, any part of the article may be reused without permission provided that the original article is clearly cited. For more information, please refer to https://www.mdpi.com/openaccess .

Feature papers represent the most advanced research with significant potential for high impact in the field. A Feature Paper should be a substantial original Article that involves several techniques or approaches, provides an outlook for future research directions and describes possible research applications.

Feature papers are submitted upon individual invitation or recommendation by the scientific editors and must receive positive feedback from the reviewers.

Editor’s Choice articles are based on recommendations by the scientific editors of MDPI journals from around the world. Editors select a small number of articles recently published in the journal that they believe will be particularly interesting to readers, or important in the respective research area. The aim is to provide a snapshot of some of the most exciting work published in the various research areas of the journal.

Original Submission Date Received: .

• Active Journals
• Find a Journal
• Proceedings Series
• For Authors
• For Reviewers
• For Editors
• For Librarians
• For Publishers
• For Societies
• For Conference Organizers
• Open Access Policy
• Institutional Open Access Program
• Special Issues Guidelines
• Editorial Process
• Research and Publication Ethics
• Article Processing Charges
• Testimonials
• Preprints.org
• SciProfiles
• Encyclopedia

Article Menu

• Subscribe SciFeed
• Recommended Articles
• Google Scholar
• on Google Scholar
• Table of Contents

Find support for a specific problem in the support section of our website.

Please let us know what you think of our products and services.

Visit our dedicated information section to learn more about MDPI.

JSmol Viewer

Evaluation of thermal stress on heterogeneous iot-based federated learning  †.

1. Introduction

• To the best of our knowledge, the present work is among the first to evaluate the performance of federated learning under thermal stress on real-world heterogeneous IoT-based systems.
• We conducted experiments of various scenarios when federated learning clients are under different thermal stresses with measuring metrics, including CPU utilization rate, GPU utilization rate, temperature, and power consumption. We added 67% more training epochs compared with our previous work [ 3 ].
• We varied the proportion of clients under thermal stress in each group of experiments and systematically quantified the effectiveness and real-world impact of thermal stress on the low-end, heterogeneous, IoT-based federated learning system. We added 50% more clients compared with our previous work [ 3 ].

2. Related Work

3. methodology, 3.1. system description, 3.2. stress implementation, 4. experimental evaluation, 4.1. evaluation methodology, 4.2. thermal stress simulation, 4.3. measurements of metrics, 4.3.1. training time and accuracy, 4.3.2. cpu and gpu utilization rate and total energy consumption, 4.3.3. temperature, 4.4. experiment process and outcomes, 5. discussion and analysis, 5.1. impact on cpu and gpu utilization rate, 5.2. temperature and power consumption, 5.3. impact on training time and accuracy, 5.4. analysis and insights, 6. conclusions, limitations and future work, 6.1. conclusions, 6.2. limitations, 6.3. future work, author contributions, data availability statement, conflicts of interest.

• McMahan, B.; Ramage, D. Federated Learning: Collaborative Machine Learning without Centralized Training Data. Available online: https://research.google/blog/federated-learning-collaborative-machine-learning-without-centralized-training-data/ (accessed on 24 June 2024).
• Gouissem, A.; Abualsaud, K.; Yaacoub, E.; Khattab, T.; Guizani, M. Federated Learning Stability Under Byzantine Attacks. In Proceedings of the 2022 IEEE Wireless Communications and Networking Conference (WCNC), Austin, TX, USA, 10–13 April 2022; pp. 572–577. [ Google Scholar ] [ CrossRef ]
• Gu, Y.; Zhao, L.; Deng, B.; Wu, S. Evaluation of Thermal Stress on IoT-based Federated Learning. In Proceedings of the 2024 ACM Southeast Conference, ACM SE ’24, Marietta, GA, USA, 18–20 April 2024; pp. 291–296. [ Google Scholar ] [ CrossRef ]
• Beutel, D.J.; Topal, T.; Mathur, A.; Qiu, X.; Fernandez-Marques, J.; Gao, Y.; Sani, L.; Li, K.H.; Parcollet, T.; de Gusmão, P.P.B.; et al. Flower: A Friendly Federated Learning Research Framework. arXiv 2020 , arXiv:2007.14390. [ Google Scholar ]
• Bhide, A.; Kulkarni, A. NVIDIA-AI-IOT Jetson_Benchmarks. Available online: https://github.com/NVIDIA-AI-IOT/jetson_benchmarks (accessed on 24 June 2024).
• King, I.C. ColinIanKing-stress-ng. Available online: https://github.com/ColinIanKing/stress-ng (accessed on 24 June 2024).
• Bonghi, R. Jetson-Stats. Available online: https://github.com/rbonghi/jetson_stats (accessed on 24 June 2024).
• Hennion, N. Nicolargo-Glances. Available online: https://github.com/nicolargo/glances (accessed on 24 June 2024).
• McMahan, B.; Moore, E.; Ramage, D.; Hampson, S.; y Arcas, B.A. Communication-efficient Learning of Deep Networks from Decentralized Data. In Proceedings of the Artificial Intelligence and Statistics, PMLR, Fort Lauderdale, FL, USA, 20–22 April 2017; pp. 1273–1282. [ Google Scholar ]
• Google. Tensorflow Federated: Machine Learning on Decentralized Data. Available online: https://www.tensorflow.org/federated (accessed on 24 June 2024).
• Bonawitz, K.; Eichner, H.; Grieskamp, W.; Huba, D.; Ingerman, A.; Ivanov, V.; Kiddon, C.; Konečný, J.; Mazzocchi, S.; McMahan, H.B.; et al. Towards Federated Learning at Scale: System Design. arXiv 2019 , arXiv:1902.01046. [ Google Scholar ]
• Ryffel, T.; Trask, A.; Dahl, M.; Wagner, B.; Mancuso, J.; Rueckert, D.; Passerat-Palmbach, J. A Generic Framework for Privacy Preserving Deep Learning. arXiv 2018 , arXiv:1811.04017. [ Google Scholar ]
• Masti, R.J.; Rai, D.; Ranganathan, A.; Müller, C.; Thiele, L.; Capkun, S. Thermal Covert Channels on Multi-core Platforms. In Proceedings of the 24th USENIX Security Symposium (USENIX Security 15), Washington, DC, USA, 12–14 August 2015; pp. 865–880. [ Google Scholar ]
• Tian, S.; Szefer, J. Temporal Thermal Covert Channels in Cloud FPGAs. In Proceedings of the 2019 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays, Seaside, CA, USA, 24–26 February 2019; pp. 298–303. [ Google Scholar ]
• Kong, J.; John, J.K.; Chung, E.Y.; Chung, S.W.; Hu, J. On the Thermal Attack in Instruction Caches. IEEE Trans. Dependable Secur. Comput. 2010 , 7 , 217–223. [ Google Scholar ] [ CrossRef ]
• Gao, X.; Xu, Z.; Wang, H.; Li, L.; Wang, X. Why “Some” Like It Hot Too: Thermal Attack on Data Centers. In Proceedings of the 2017 ACM SIGMETRICS/International Conference on Measurement and Modeling of Computer Systems, Urbana-Champaign, IL, USA, 5–9 June 2017; pp. 23–24. [ Google Scholar ]
• Gao, X.; Xu, Z.; Wang, H.; Li, L.; Wang, X. Reduced Cooling Redundancy: A New Security Vulnerability in a Hot Data Center. In Proceedings of the NDSS, San Diego, CA, USA, 18–21 February 2018. [ Google Scholar ]
• Duchatellier, J.; Holmes, T.; Suo, K.; Shi, Y. An Empirical Study of Thermal Attacks on Edge Platforms. In Proceedings of the 2021 ACM Southeast Conference, New York, NY, USA, 15–17 April 2021; pp. 175–179. [ Google Scholar ]
• Kaur, J.; Das, S. RSPP: Restricted Static Pseudo-Partitioning for Mitigation of Cross-Core Covert Channel Attacks. ACM Trans. Des. Autom. Electron. Syst. 2024 , 29 , 1–22. [ Google Scholar ] [ CrossRef ]
• Afermarq; Tanertopal. Federated Learning on Embedded Devices with Flower. arXiv 2023 , arXiv:2007.14390. [ Google Scholar ]
• Paszke, A.; Gross, S.; Massa, F.; Lerer, A.; Bradbury, J.; Chanan, G.; Killeen, T.; Lin, Z.; Gimelshein, N.; Antiga, L.; et al. PyTorch: An Imperative Style, High-Performance Deep Learning Library. In Proceedings of the 33rd International Conference on Neural Information Processing Systems ; Curran Associates Inc.: Red Hook, NY, USA, 2019. [ Google Scholar ]
• LeCun, Y.; Cortes, C. MNIST Handwritten Digit Database. Available online: https://yann.lecun.com/exdb/mnist/ (accessed on 24 June 2024).
• Svcavallar. Jetson-Stats-Grafana-Dashboard. Available online: https://github.com/svcavallar/jetson-stats-grafana-dashboard (accessed on 24 June 2024).
• Authors, P. Prometheus Monitoring Systems and Time Series Database. Available online: https://prometheus.io/ (accessed on 24 June 2024).
• Labs, G. Grafana: The Open Observability Platform. Available online: https://grafana.com/ (accessed on 24 June 2024).
• Krizhevsky, A.; Nair, V.; Hinton, G. CIFAR-10 (Canadian Institute for Advanced Research). Available online: https://www.cs.toronto.edu/~kriz/cifar.html (accessed on 24 June 2024).
• Saridou, B.; Bendiab, G.; Shiaeles, S.N.; Papadopoulos, B.K. Thermal Management in Large Data Centres: Security Threats and Mitigation. In Proceedings of the Security in Computing and Communications: 8th International Symposium, SSCC 2020, Chennai, India, 14–17 October 2020; Revised Selected Papers 8. Springer: Chennai, India, 2021; pp. 165–179. [ Google Scholar ]
• Hou, X.; Wang, J.; Jiang, C.; Meng, Z.; Chen, J.; Ren, Y. Efficient federated learning for metaverse via dynamic user selection, gradient quantization and resource allocation. IEEE J. Sel. Areas Commun. 2023 , 42 , 850–866. [ Google Scholar ] [ CrossRef ]
• Amiri, M.M.; Gündüz, D. Federated learning over wireless fading channels. IEEE Trans. Wirel. Commun. 2020 , 19 , 3546–3557. [ Google Scholar ] [ CrossRef ]

Click here to enlarge figure

Share and Cite

Gu, Y.; Zhao, L.; Liu, T.; Wu, S. Evaluation of Thermal Stress on Heterogeneous IoT-Based Federated Learning. Electronics 2024 , 13 , 3140. https://doi.org/10.3390/electronics13163140

Gu Y, Zhao L, Liu T, Wu S. Evaluation of Thermal Stress on Heterogeneous IoT-Based Federated Learning. Electronics . 2024; 13(16):3140. https://doi.org/10.3390/electronics13163140

Gu, Yi, Liang Zhao, Tianze Liu, and Shaoen Wu. 2024. "Evaluation of Thermal Stress on Heterogeneous IoT-Based Federated Learning" Electronics 13, no. 16: 3140. https://doi.org/10.3390/electronics13163140

Article Metrics

Article access statistics, further information, mdpi initiatives, follow mdpi.

Subscribe to receive issue release notifications and newsletters from MDPI journals