Contents I started my journey in artificial intelligence (AI) long ago, in 1994, after a summer course at MIT by Prof. Jean-Jacques Slotine, director of the Nonlinear Systems Laboratory. According to early references, AI applied to power systems was already considered mainstream. Although fuzzy logic, pattern recognition, and expert systems were discussed, many of these investigations were based on the Rosenblatt perceptron, invented by Frank Rosenblatt in 1957, one of the earliest models in machine learning (ML). However, the perceptron was notoriously hard to train and could solve only simple classification tasks. The neural nets awoke from this dormancy in the mid-nineties after enhanced training algorithms unlocked new possibilities, especially in image classification. The “backpropagation” proposed by David E. Rumelhart, Geoffrey E. Hinton, and Ronald J. Williams solved training limitations for a larger class of classification problems using multilayer perceptron. Interestingly, G. E. Hinton is a 2024 Nobel Prize Recipient in Physics for “foundational discoveries and inventions that enable machine learning with artificial neural network.”
The potential of neural networks (NNs) in hard multiclass classification tasks, such as fault-type identification and instability screening, suddenly attracted power systems engineers. The latter is harder than naively thought because 90%+ of cases are typically stable. At the same time, the engineer wants to classify 100% of all small subsets of dangerous and unstable cases (after all, who wants to be responsible for a blackout?). NNs were tried with some success on these skewed classification problems but with mixed success (see Yakout Mansour et al. for lessons learned from project reports for the Canadian Association for Electricity between 1993 and 1996). Some success was also reported in applying artificial NNs to load forecasting in energy management systems (EMSs) in New York State.
But frankly, AI did not deliver on business value as quickly as was needed, and its prestige faded away. The reasons were numerous. The usual suspects were again the training, which was too slow, considering the computing power and the learning algorithm’s limitations. This was compounded by the difficulty of creating datasets suitable for the learning problems. There is a fascinating 1997 article by T. E. Dy-Liacco in IEEE Computer Applications in Power, the ancestor of this magazine, which discussed how to generate the data required for creating an AI-based “automatic operator” for enhancing the power system security control. Such an application was actively investigated because of the perception that since automatic learning methods were computationally efficient, they could replace analytical procedures for security analysis, thus enabling the dream of the “real-time implementation” of such schemes. The article argued rightly that all automatic learning approaches to system security analysis or other network-oriented problems are confronted by the obstacle of training-set generation arising from the tremendously large number of combinations of variables and topologies and the computation required for each simulation.
Now, even if we resolve this huge issue of the data training dataset for a system that stays quiet without any problem 99.9% of the time, what is the added value for doing that? Is it worth the trouble of dealing with the inseparable baggage of training-set generation and maintenance just to use a slightly more efficient tool or set of tools for security analysis? Dy-Liacco’s answer to this “value-added” question is yes if the system is large and complex enough to override the cognitive capability of a human operator. Hence, AI-based automatic operators can go beyond secure/insecure binary answers to provide for medium to large meshed networks with more refined operator guidance. A graduated security or severity index would be more appropriate than a strict dichotomy. In addition, AI-based identification of the attributes strongly associated with or influencing the severity should be made available to the operator; such attributes should include the worst offending contingencies. To help identify the options for preventive control, high generation levels, topology weaknesses, and unsafe branch loadings could be among the key attributes. These brilliant ideas were already around in 1997. But still, no noticeable action was noticed in utilities toward capturing the added value in shifting control from a human-centric top-down system to a more automated data-driven system of systems.
Today, we are witnessing a renewed interest in using AI to solve power systems problems (Figure 1). According to the KPMG 2023 CEO Outlook (https://kpmg.com/xx/en/what-we-do/industries/energy-natural-resources-chemicals/power-utilities/plugged-in-magazine-3/how-artificial-intelligence-and-automation-can-help-transform-power-and-utilities.html), 64% of energy CEOs agree that investing in generative AI is a top priority, with 48% expecting a return on their investment in three to five years. What has changed between 1997 and now? There are many possible explanations, although the biggest and most obvious may be the enhancement of the AI technologies through efficient hardware (e.g., from NVIDIA), and software (ChatGPT, deep learning, etc.). Simply put, AI is everywhere in the news because the underlying technologies have become mature, cheaper, and more effective in addressing our day-to-day problems without requiring too much expert involvement as well as less training data on our part. Since 2010, the computational power needed to develop increasingly sophisticated ML models has doubled approximately every five to six months. Today, AI models can accurately perform language and image recognition tasks, converting audio into analyzable data, powering chatbots, and automating simple tasks. By analyzing data and inputs, AI replicates certain aspects of human intelligence, producing outputs faster and in greater quantities than a human could. Some AI algorithms can even self-program and alter their code.
Electricity highways in synergy with AI highways. (Graphic(s) created using AI generation. For image credits, please see the “Acknowledgment” section of this article.)
Alongside the enhancements in AI technologies, power systems have grown in complexity as electricity demand increases and decarbonization efforts intensify. As Dy-Liacco anticipated, it is more difficult for engineers to tame them using human cognitive capabilities only. Indeed, traditional grids channeled energy from centralized power stations, but nowadays, they must accommodate multidirectional electricity flows among distributed generators, the grid, and users. The growing number of grid-connected devices, such as electric vehicle (EV) charging stations and residential solar panels, adds unpredictability to these flows. The power system is also becoming more interconnected with the transportation and building industrial sectors. This evolution necessitates a much greater exchange of information and more advanced tools for planning and operating power systems. Here is why AI comes into play.
Unsurprisingly, the energy sector is leveraging AI to enhance efficiency and drive innovation under decarbonization and digitization transitions. AI is particularly well suited to support the growth of smart grids and manage the vast amounts of data they produce. Smart meters generate and transmit thousands of times more data points to utilities compared to their analog counterparts. New grid power flow monitoring devices send significantly more data to operators than the older technologies they replace. Additionally, the global fleet of digital-aware wind turbines is estimated to generate more than 400 billion data points annually in the global fleet of wind (https://www.evwind.es/2020/05/01/the-worlds-wind-turbines-are-registering-more-than-400-billion-individual-data-points-every-year/74591). Phasor measurement units (PMU) are streaming even more data to EMSs, and the shift toward synchronized waveform measurement units will multiply the data volume by 256+. This sheer volume of data is a significant reason why energy companies view AI as an increasingly vital resource. Recent estimates indicate that AI is already being used in more than 50 different applications within the energy system. Additionally, the market size for AI in energy and utilities (https://www.industryarc.com/Report/17916/artificial-intelligence-market-in-energy-utilities.html) is expected to expand to US
At the same time, AI consumes more energy than other forms of computing, which is an important consideration as the world aims to create a more efficient energy system. Training a single AI model uses more electricity than 100 U.S. homes consume annually. In 2022, Google reported that ML accounted for about 15% of its energy use over the previous three years. Your Google search and ChatGPT queries consume ¯0.3 Wh and 2.9 Wh, respectively. Hence, it is estimated that 33% of electricity consumption in Ireland will be from data centers.
One of the primary applications of AI in the energy sector is facilitating demand response and maximizing online renewable energy through more accurate forecasting of electricity demand, supply, and prices, scheduling and managing response loads, and implementing dynamic pricing (Figure 2). Wind power output can be predicted by using weather models and data on turbine locations. Google and its AI subsidiary DeepMind (https://blog.google/technology/ai/machine-learning-can-boost-value-wind-energy/) developed an NN in 2019 to improve forecast accuracy for their 700-MW renewable energy fleet. Using historical data, the network created a model to predict future output up to 36 h in advance with significantly greater precision than before. This enhanced visibility enables Google to sell its power in advance rather than in real time. The company has reported that this, along with other AI-driven efficiencies, has boosted the financial value of its wind power operations by 20%.
The innovation potential of AI stretches into innumerable power systems applications. However, only a few of them are mature enough for implementation. V2G: vehicle-to-grid; DoS: denial-of-service; FDI: false data injection.
Another significant AI application is predictive maintenance (https://micro.ai/resources/case-studies/enabling-predictive-maintenance-in-energy-production), which involves continuously monitoring and analyzing the performance of energy assets to detect potential faults before they occur. Two articles in this issue will illustrate AI use cases for these problems. Traditionally, maintenance is performed on a fixed schedule; for example, transmission line poles might be inspected periodically, with repairs made as necessary. This uniform approach can be inefficient if maintenance is done too early or, more critically, too late. To improve this, various utilities are developing AI-driven programs to monitor physical assets and use historical performance and outage data to predict when maintenance is needed. Two articles in this issue will illustrate AI use cases for these problems. Similarly, the utility company E.ON (https://www.eon.com/en/new-energy/digitization/artificial-intelligence.html) has created an ML algorithm to forecast when medium-voltage cables in the grid need replacement, utilizing data from multiple sources to identify patterns in electricity generation and highlight any irregularities. E.ON’s research indicates that predictive maintenance could reduce grid outages by up to 30% when compared to traditional methods. Another application that is still in the research stage but with great promise is managing and controlling grids; this involves utilizing data from sensors, smart meters, and other Internet of Things devices to monitor and regulate power flow within the network, especially at the distribution level. Furthermore, companies like Octopus Energy and Oracle Utilities are also exploring how to enhance consumer services using AI or ML in apps and online chatbots to improve customer billing experiences.
Without AI, system operators and utilities can utilize only a small portion of emerging digital technologies’ new data sources and processes, missing out on many potential benefits. However, addressing associated risks, such as cybersecurity and privacy threats, data biases or errors, and miscorrelations due to inadequate training data or coding mistakes is crucial before scaling AI across the sector. Additionally, the increased use of automated and self-learning software raises questions about accountability for the outputs or outcomes of these systems. Operators often purchase AI technology or related services from IT companies and startups, which can lead to decision making on electricity balancing or investments based on models that they do not fully understand or control (Figure 3). This situation raises concerns about accountability for public spending, energy prices, and outages. To address these challenges, the Organization for Economic Cooperation and Development (OECD) AI Principles (https://www.europarl.europa.eu/news/en/press-room/20230505IPR84904/ai-act-a-step-closer-to-the-first-rules-on-artificial-intelligence), adopted in 2019 by OECD members, offer guidance on a human-centric approach to trustworthy AI. Given the energy sector’s critical role in the global economy and climate goals, clearer national, regional, and international frameworks may be necessary.
For casual users, AI is a “black box.” What software is inside? Who is responsible for the design flaws?
The aforementioned context sets the stage for this issue on AI. During the 2024 presidential debate, Harris stressed the need to invest in “American-based technology” so that the United States could win “the competition for the 21st century” or the “race on A.I. and quantum computing.” She added, “I will make sure that we lead the world into the future on space and artificial intelligence.” There are already many ongoing activities in North America, China, and Europe with the goal of understanding AI roles and its benefits for power systems; a further consideration is also that of alleviating the barriers preventing wider adoption of effective AI-based solutions in the electricity utility sector provided that AI innocuity and this critical infrastructure’s reliability and security are convincingly demonstrated. Hence, the International Energy Agency (https://www.iea.org/commentaries/why-ai-and-energy-are-the-new-power-couple) published a 2023 white paper on the ineluctable convergence between AI and energy. At the same time, the U.S. Department of Energy (DOE) published, in April 2024, a report, “AI for Energy: Opportunities for a Modern Grid and Clean Energy Economy.” It outlines the DOE vision of leveraging AI to speed the race toward a 100% clean energy system. IEEE has also established a Task Force on Machine Learning for Power Systems and a Working Group on Big Data Analytics for Power Systems. Not to be left behind, the IEEE Power & Energy Magazine 2022 March/April issue (https://ieeexplore.ieee.org/document/9761144) was dedicated to ML applications in power systems. Therefore, we are not building on an empty field but rather complementing the rich ecosystem of AI for power systems resources available to our readers through various channels.
The issue, for once, is an editorial board pull, not a guest editor push. In the spring of 2023, Prof. Chongqing Kang, the dean of Electrical Engineering at Tsinghua University, an IEEE Fellow, and a widely acclaimed expert on AI applications to power systems, was invited to lead the proposal of a special issue on AI meeting certain requirements mandated by IEEE Power & Energy Society (PES) regulations. First, the magazine had just published a special issue a year before, and we sought to avoid any duplication of content. Second, we wanted to discuss why AI is not more pervasive in our industry. One may argue, as shown before, that there are a lot of activities out there surrounding AI applications in power systems. A search in IEEE Xplore (https://ieeexplore.ieee.org/search/searchresult.jsp?newsearch=true&queryText=Artificial%20intelligence%20and%20power%20systems%E2%80%9D) of the keywords “artificial intelligence and power systems” at the time of this writing in the fall of 2024 returned 27,409 items of all kinds since 1910, with more than 70% published over only the last five years. This is a massive show of strength for this topic, justifying why we have another issue; it is simply at the overwhelming request of the community. With this, the result is a nicely structured issue with articles from around the world.
Further details regarding the nine feature articles are provided in “In This Issue.” They cover the usual segments of AI applications found in power systems, namely, deep learning-based resources forecasting; AI in combination with data-intensive technologies, such as virtual power plants (VPPs), wide-area measurement systems (WAMSs), and smart metering infrastructure with consideration of data privacy and cybersecurity issues; reinforcement learning for networked microgrids; and AI for proactive asset management, fault detection and outage management in hydropower plants, transmission, and distribution power networks. Two further articles touch upon learning-based modeling and security monitoring and control within a control center environment, as envisioned by T. E. Dy-Liacco two decades ago. At the same time, the “In My View” column [A1] thoughtfully discusses why utilities are reluctant to adopt AI-based solutions to their business problems. It also suggests ways to alleviate barriers through collaboration, particularly in using and sharing power system data to optimize systems operations (Figure 4).
Increased computational power and hyperscale cloud resources have transformed our previous assumptions about AI’s potential to learn, execute (https://www.youtube.com/watch?v=OlPNVqlhon8), and optimize the smart power grid (https://www.utilitydive.com/news/nvidia-ibm-schneider-utilidata-ai-utility-regulation-grid-distribution/716538/), from planning to operations, asset management, and customer service—starting utilities on what some are calling the “technology transition.” (Source: U.S. Energy Association.)
In This Issue
Power systems around the world are becoming increasingly more complex, with the extensive use of established and emerging technologies that are often based on power electronics-interfaced devices deployed in power generation, transmission and distribution networks, and consumer sides. As the sheer number of these devices soars in orders of magnitude, their complexity and flexibility also increase. This warrants the need for the higher quality and quantity of the data than has been required to date for secure and reliable power system planning, design, and operation. IT systems have been evolving in the past few years to better cater to this increasing need as the journey to energy transition advances.
Faster data processing and simulation speeds for power system studies have become possible, and many other initiatives are underway around the world. However, the need to better understand systems that are more complex also creates a need to more accurately model the power system with additional details, sometimes with a much higher impact on the computational burden. Therefore, while IT systems combined with contemporary power systems, simulation programs, and techniques will serve as key enablers to facilitate a secure, timely, and efficient energy transition, there remain some gaps with respect to the need for faster data processing and, importantly, more confident and automated decision making in various aspects of power systems relying on the higher quantity and quality of data and models. This is where AI and ML are emerging in various power system applications, as discussed in this issue of IEEE Power & Energy Magazine.
The first four articles are more focused on distribution systems, microgrids, and distributed energy resources, although most principles and, in particular, generation and load forecasting and data sharing and security, are highly relevant for transmission systems and the whole system performance. These four articles cover the following:
VPPs are progressively playing a key role in coordinating and orchestrating several hundreds if not thousands of distributed energy resources, such as distributed photovoltaics (PVs) and EVs. Noting the sheer number of the devices involved, which often exhibit inferior controllability and visibility compared to utility-scale transmission-connected inverter-based resources, [A2] discusses the applications of AI with better analytical and “thinking” capability for several VPPs in China. The application of various AI technologies is discussed, and three use cases classified as edge intelligence analysis, cloud complex decision making, and cloud-edge collaborations are presented.
With increasing amounts of data available from smart meters, it is prudent to share and use these data at a coordinated and whole-system level rather than for each individual device. AI can help analyze and make the best use of the high volume of the data involved. However, the privacy and security of the shared data will become more important than ever, as pointed out in [A3], where various methods are discussed that could help preserve data security, together with various case studies that utilize AI applications.
One of the key applications of AI in power systems to date has been for load and generation forecasting, which is an important aspect of ensuring adequate resource allocation and correct system dispatch. This topic is covered in [A4], which provides extensive background information on the key technologies used, such as deep reinforcement learning (DRL) and knowledge-infused-DRL. The article then discusses practical examples of using ML for real-time power system operation in a provincial power system and a distribution system.
The application of AI-driven reinforcement learning in multi-microgrid power systems, intended to increase system resilience and stability under abnormal operating conditions such as extreme weather events, is discussed in [A5]. A key objective is the optimal and coordinated control of several otherwise decentralized microgrids with the use of AI to tackle the uncertainties associated with the events and to allow a fast dynamic response in the order of milliseconds without the need to use optimization techniques. A performance comparison of AI-and optimization-based approaches is presented demonstrating sufficient accuracy achieved with the AI-based method.
Another major sector in the power industry is power generation, which can also benefit from AI. There is one article in this issue focusing on AI applications for monitoring and diagnosis of faults and defects in large hydrogenerators [A6]. The advantages of AI-based methods are demonstrated compared to the conventional methods with two case studies associated with operational hydrogenerators. These advantages include the earlier identification of a potential fault, automated decision making, and the ability to deal with multidimensional inputs, such as those containing both time-domain and frequency-domain information.
The remaining articles will focus more on bulk power systems, though one of the articles also discusses the application of AI for outage management in distribution systems.
Article [A7] presents extensive discussions on the use of AI and ML in power system protection with a particular focus on fault analysis in transmission networks and outage management in the distribution networks. The key objective in fault analysis is to automate analyzing the results and detecting and classifying the faults with the use of more data from PMUs. For distribution system outage management, the use of ML allows for predicting the state of the risk of outages by gathering information from historical outages caused by various environmental impacts.
Article [A8] provides an overview of various AI and ML applications in the South Korean power system, including 1) a WAMS for improved fault classification; 2) an EMS for reducing the processing time for system stability analysis; 3) ML-based data-driven dynamic models and model parameters identification; and 4) detecting abnormal conditions and re-establishing control under such conditions for high-voltage dc (HVdc) systems. Among the key problems aimed to be addressed with these developments is mitigating the frequent curtailment of solar PV generation recently experienced in South Korea.
Similar to [A8], article [A9] focuses on the application of AI and ML for bulk power system dynamic modeling and analysis. Practical examples are presented that describe how transmission system operators currently use them in real-time power system operation to complement other existing and emerging tools and techniques to make a confident decision about the power system’s dynamic performance. The article also shares insights on both the opportunities and limitations of AI and ML in power system dynamic analysis as well as a prognosis of the way forward. The article includes excellent background information, which can also aid in better understanding some of the principles and concepts discussed in other articles.
Touching on the significant complexity and dynamic interdependency between local and national grid systems as the power systems are transitioning to a net-zero energy future, the “In My View” column [A1] discusses the limited economic and operational benefit realized from the deployment of AI in practical industrial environments, such as electrical power systems. As indicated in the article, unlocking the value of AI within power systems to facilitate autonomous control, operation, and monitoring capability remains a key challenge. It is also highlighted that we are moving to a world where power systems will require engineers and innovators to be as fluent with data, AI tools, and technologies as they are with electrical theories. The article also cautions about the risk of becoming overly fixated on novel techniques and not the effective use of them. Critical questions to be addressed in this regard include:1) determining the operational and business benefits of AI; 2) building trustworthiness, safety, and reliability in the AI used within the power system; and 3) determining the balance between autonomy and human control.
—Babak Badrzadeh, Associate Editor, Issues
Leader’s Corner
Jackie Peer, the PES vice president of conferences and meetings, authored the “Leader’s Corner” column [A10] in this last issue of 2024. Despite her heavy schedule, we greatly appreciate her contributing to the magazine to benefit our readers. The article regarding the IEEE conference program track record and prospects is very well documented. Amazingly, our flagship conference, the PES General Meeting, held in Seattle in July 2024, has attracted a record number of registrations at 3,538, representing 61 countries, presenting 686 conference papers and 178 transaction articles. Not only is our conference program stellar in attendance and technical activities, but it is also notable for its diversity, equity, and inclusion achievement, considering the regional, gender, and racial diversity of participants from around the globe. As a side note, the PES General Meeting has a high rejection rate, resulting in only high-quality papers being presented there; this has led to the proceedings of the conference being ranked by Google Scholar (https://scholar.google.com/citations?view_op=top_venues&hl=en&vq=eng_powerengineering) above many power engineering journals in terms of its scholarly impact. So, I encourage you to join us as presenters or curious participants eager to learn what’s new in power and energy at the 2025 PES General Meeting in Austin, TX, USA.
History Column
This month, our “History” column [A11] welcomes first-time contributor Richard B. Miller, a graduate of Amherst College and New York University School of Law. Richard explores the characteristics of Edison’s Pearl Street Station in Manhattan, NY, USA, as the first power station in the world to power a commercial neighborhood electric distribution network designed to serve all users. We are now approaching the 150th anniversary in 2032, which makes the case as the origin of today’s electric “grid.”
Magazine News Desktop
We have no newsworthy information except for recent changes in the PES Magazines organization. IEEE Electrification Magazine (IEM) has undergone a big transformation; the major sponsoring Society of IEM has been changed from PES to the Transportation Electrification Council, and the official memorandum of understanding was signed in June 2024. The financial share among TEC/PES/PELS/IAS is now 55%/15%/15%/15%. The steering committee met online on 8 July 2024 and approved the appointment of the new editor-in-chief (EIC), Prof. Yaobin Chen of Purdue University (starting date: 1 January 2025). We take this opportunity to welcome Yaobin to this new role and express our sincere thanks for the many years of dedicated service from the outgoing EIC, Prof. Lingling Fan, of the University of South Florida. Lingling, your good work has not gone unnoticed within the PES community.
On the other side, PES has launched a new magazine, IEEE Energy Sustainability Magazine (https://www.ieee.org/membership-catalog/productdetail/showProductDetailPage.html?product=PER379-ELE), fully financially sponsored by PES and to be published quarterly (print and electronic versions), with the first issue planned in May 2025. Prof. Mohammad Shahidehpour has been appointed the inaugural/founding EIC of IEEE Energy Sustainability Magazine. He was also the founding EIC of IEEE Transactions on Smart Grid. Congratulations, Prof. Shahidehpour, for your long-lasting contribution to PES technical activities.
Wrap-Up
This section tends to be repetitive, yet I have no choice but to close my editorial by thanking all the volunteers who made this issue possible through their around-the-clock work. First and foremost, thanks go to our guest editors, Chongqing Kang (Tsinghua University, China), Ricardo Bessa (INESC TEC, Portugal), and Yi Wang (The University of Hong Kong, Hong Kong, China). We also sincerely appreciate the many authors who graciously accepted our invitation to participate amid their hectic schedules. The proficiency of the editors and authors in this rapidly cutting-edge research field is truly exceptional, and we would like to extend our hearty commendation to them. It has not gone unnoticed that almost all the authors are from academia, which is unusual for IEEE Power & Energy Magazine. The reason can be easily understood as the field is nascent. Nevertheless, some articles involve utility research centers in Canada and Korea. While it is somehow normal that academia is at the front stage today, given the lack of utility implementation of the newest AI technologies, we commit that our next issue on AI in power systems will be approved only if it is industry driven. We encourage the industry to jump in with a proposal aligned with their perspectives. Regarding the scientific content, the tireless work of Dr. Babak Badrzadeh, our reliable associate editor, in proofreading the initial versions is very much appreciated. His experience in magazine article writing, the constant support to the authors, and timeliness at all stages, including the “In This Issue” contribution to this editorial, were much appreciated.
In this issue, we also have two unsolicited articles that are not directly related to AI. They warrant placement in this issue, however, because they provide the foundation for power systems to host AI applications, a sequel to this year’s May/June issue on the digital grid. The first article, addressing digitization [A12], is from Brazil and was contributed by coauthors from the Polytechnic School of the University of São Paulo and one of the major transmission companies in Brazil, ISA CTEEP, also based in São Paolo. Titled “Bridging Industry 4.0 and Power Systems: A Conceptual Framework for the Substation of the Future,” the article aimed at defining the concept of Substation 4.0 and discussed how it is being embraced by transmission operators struggling to cope with rapid energy transition in a digital era. The concepts involved in the idea of Substation 4.0 are organized into pillars, enablers, and technologies, which are presented along with the definitions adopted for each one of them. The authors explain how improving the operational excellence of the system can be accelerated by Industry 4.0 technologies and concepts such as digital transformation and process automation. Allowing the fast and efficient acquisition of a large amount of data and making them available to operators and specialists can contribute in a relevant way to better-informed design decisions, failure prediction, and operation alternatives with the help of AI tools, business intelligence, and other useful analyses.
The second article addresses data communications in the grid edge [A13] from the Queensland University of Technology, Brisbane, Queensland, Australia; Swinburne University of Technology, Melbourne, Australia; and East China University of Science and Technology, Shanghai, China. The article’s focus is the Neighborhood Area Networks (NANs), which connect all communication devices between the substations and smart meters. This constitutes the fundamental last-mile infrastructure for controlling electricity distribution networks. By mapping smart grid functionalities to the data communication models compounded by a survey on smart grid communication standards relevant to NANs, the authors clarify the standards in use and the associated protocols suitable for NAN implementation. Furthermore, they identified some challenges worthy of future investigation.
Given the political, environmental, and social aspects of AI, the issues related to data privacy, and the need for more responsible use of AI and enhanced energy efficiency in AI, we are very interested in reading your letters in response to the opinions expressed in the editorials and articles published on this issue. We are especially interested in AI implementation and adoption cases in the energy sector as well as supporting business cases and lessons learned. Publishing a letter to the editor gives a voice to the people this magazine serves. You can attach your letter as a Word file to powerandenergymagazine@gmail.com.
Visit our website to learn more about future issues up to the first half of 2025. On this occasion, I invite you to suggest subjects within the scope of this magazine that you would like us to address. You can e-mail me at innocent.kamwa@gel.ulaval.ca. Since the planning for 2025 is closed, we have started discussing new topics to be featured in the magazine in 2026 and beyond. Also, for those who want to use the magazine’s reach (distributed to all 40,000+ PES members) to promote their product, please don’t forget that you can place an ad in each issue by contacting our ad representative, Naylor (see the back cover of the magazine for more information). We are also open to publishing paid sponsored content for institutions that prefer this approach to reach their target audience. Finally, if you have not noticed yet, the present piece, called “Editors Voice,” is open access, meaning that thousands of readers, not only PES members, can see each ad in this article over an indefinitely long period of time.
ACKNOWLEDGMENT
Figure 1 was generated by ChatGPT. The figure was selected among five options proposed as result of the image querie “AI in power systems.”
Appendix: Related Articles
Appendix: Related Articles
J. McDonald, “Transition to net zero: Using data-centric engineering solutions [In My View],” IEEE Power Energy Mag., vol. 22, no. 6, p. 144, Nov./Dec. 2024, doi: 10.1109/MPE.2024.3422154.
Q. Chen, X. Wang, C. Feng, C. Li, and K. Zheng, “Empowering the grid edge to think: Applications of artificial intelligence for virtual power plants in China,” IEEE Power Energy Mag., vol. 22, no. 6, pp. 66–77, Nov./Dec. 2024, doi: 10.1109/MPE.2024.3398570.
R. Yao, J. Song, Z. Li, H. Yu, and Y. Wang, “Smart meter data sharing for AI-enhanced energy systems: A review of relevant techniques and detailed case studies,” IEEE Power Energy Mag., vol. 22, no. 6, pp. 42–53, Nov./Dec. 2024, doi: 10.1109/MPE.2024.3417239.
D. Shi et al., “Toward artificial-intelligence-empowered smarter power grid: Forecasting, dispatch, and control,” IEEE Power Energy Mag., vol. 22, no. 6, pp. 54–65, Nov./Dec. 2024, doi: 10.1109/MPE.2024.3408111.
D. Qiu et al., “Artificial intelligence for microgrid resilience: A data-driven and model-free approach,” IEEE Power Energy Mag., vol. 22, no. 6, pp. 18–27, Nov./Dec. 2024, doi: 10.1109/MPE.2024.3405893.
H. Bechara et al., “Unleashing artificial intelligence: Monitoring and diagnosing large hydrogenerators,” IEEE Power Energy Mag., vol. 22, no. 6, pp. 89–99, Nov./Dec. 2024, doi: 10.1109/MPE.2024.3375250.
M. Kezunovic, R. Baembitov, and T. Mohamed, “No silver bullet: Artificial intelligence is not a panacea, but it works for fault analysis and outage management,” IEEE Power Energy Mag., vol. 22, no. 6, pp. 78–88, Nov./Dec. 2024, doi: 10.1109/MPE.2024.3415568.
J. Lee et al., “South Korean power system operation and renewable integration: Using artificial intelligence applications,” IEEE Power Energy Mag., vol. 22, no. 6, pp. 28–41, Nov./Dec. 2024, doi: 10.1109/MPE.2024.3411580.
P. N. Papadopoulos, S. Chatzivasileiadis, and A. Marot, “Can machine learning help keep the system secure? Power systems and change addressing the increasing complexity and uncertainty during the energy transition,” IEEE Power Energy Mag., vol. 22, no. 6, pp. 100–111, Nov./Dec. 2024, doi: 10.1109/MPE.2024.3421388.
J. Peer, “PES events: reflecting on some of the highlights [Leader’s Corner],” IEEE Power Energy Mag., vol. 22, no. 6, p. 15, Nov./Dec. 2024, doi: 10.1109/MPE.2024.3475409.
R. B. Miller, “Pearl street station: powering the first distribution network [History],” IEEE Power Energy Mag., vol. 22, no. 6, pp. 134–140, Nov./Dec. 2024, doi: 10.1109/MPE.2024.3464568.
G. Manassero Jr. et al., “Bridging industry 4.0 and power systems: A conceptual framework,” IEEE Power Energy Mag., vol. 22, no. 6, pp. 112–117, Nov./Dec. 2024, doi: 10.1109/MPE.2024.3422919.
B. Goswami, Y.-C. Tian, Y. Mishra, J. Jin, and Y. Tang, “Communication solutions for the last mile of smart grid: Neighborhood area networks in smart grid communications: standards and challenges,” IEEE Power Energy Mag., vol. 22, no. 6, pp. 118–133, Nov./Dec. 2024, doi: 10.1109/MPE.2024.3386652.