ipolytechiPolytech JournaliPolytech Journal2782-40042782-6341Irkutsk National Research Technical University10.21285/1814-3520-2026-1-113-124ILOIOOipolytech-1036Research ArticleЭНЕРГЕТИКАPOWER ENGINEERINGИскусственный интеллект в задачах управления энергосистемой с малой распределённой генерациейArtificial intelligence for distributed power system managementhttps://orcid.org/0009-0001-2741-6968ПрибыльскийИ. В.PribylskiiI. V.

Прибыльский Илья Викторович, аспирант

630073, г. Новосибирск, пр. К. Маркса, 20

Ilia V. Pribylskii, Postgraduate Student

20 K. Marx pr., Novosibirsk 630073 

pribilvik@mail.ruhttps://orcid.org/0000-0002-2128-9590АрмеевД. В.ArmeevD. V.

Армеев Денис Владимирович, кандидат технических наук, доцент, доцент кафедры автоматизированных электроэнергетических систем

630073, г. Новосибирск, пр. К. Маркса, 20

Denis V. Armeev, Cand. Sci. (Eng.), Associate Professor, Associate Professor of the Department of Automated Electric Power Systems

20 K. Marx pr., Novosibirsk 630073 

armeevdv@yandex.ruНовосибирский государственный технический университетРоссияNovosibirsk State Technical UniversityRussian Federation202629032026301113124Copyright © Прибыльский И.В., Армеев Д.В., 20262026Прибыльский И.В., Армеев Д.В.Pribylskii I.V., Armeev D.V.Данная работа распространяется под лицензией Creative Commons Attribution 4.0.This work is licensed under a Creative Commons Attribution 4.0 License.https://ipolytech.elpub.ru/jour/article/view/1036

Целью работы является разработка структуры и алгоритмов управления энергосистемой с распределенной генерацией на основе искусственного интеллекта, обладающей гибкостью в задании и изменении логики управления. В качестве центра принятия решений для выдачи управляющих воздействий используется большая языковая модель, применяются методы промт-инжиниринга и программирования на языках Python, SQL и Matlab. Для ускорения выполнения расчетов и снижения необходимого количества передаваемой информации в предлагаемой структуре задействуются элементы теории нечетких множеств, что позволяет использовать в алгоритме управления большую языковую модель. Разработан алгоритм проверки решений модели, позволяющий снизить вероятность ее ошибок за счет повторных независимых обработок результатов работы искусственного интеллекта и выявления несовпадений. Показано, что данный подход позволил преодолеть известную проблему, возникающую при применении технологии обработки естественного языка, связанную с непостоянством ответов. Предложенная в работе структура проверяется и оценивается с помощью серии испытаний на модели энергосистемы с малой распределенной генерацией по пяти сценариям: переход из островного режима на параллельную работу с внешней системой и наоборот; переход от сбалансированной изолированной работы двух секций на собственную нагрузку к изолированной же, но не сбалансированной по первой секции; переход от сбалансированной изолированной работы секций на собственную нагрузку к изолированной синхронной работе секций. Перечисленные испытания продемонстрировали гибкость разрабатываемой системы, ее способность принимать во внимание команды пользователя и адаптироваться к различным режимам энергосистемы. Таким образом, результаты исследования иллюстрируют возможность управления энергосистемой с малой распределенной генерацией на основе генеративного искусственного интеллекта. Возможность принятия решений и адаптивность к изменяющимся условиям делают предлагаемую систему ценным инструментом для обеспечения эффективной работы сети.

The work aims to develop a structure and algorithms for the management of a distributed generation power system based on artificial intelligence, which has the flexibility to specify and change the control logic. A large language model was used to represent a decision center for issuing control actions using Python, SQL, and Matlab prom engineering and programming techniques. In order to speed up the calculations and reduce the amount of transmitted information, the proposed structure uses elements of the fuzzy-set theory for applying a large language model in the control algorithm. The model validation algorithm was developed to reduce errors by independently reprocessing AI results and identifying mismatches. This approach was demonstrated to overcome the known problem arising from the use of the natural language processing technology associated with inconsistency of responses. The proposed structure is tested and evaluated on a distributed generation power system model in five scenarios: switching from the island mode to parallel operation with the external system and vice versa; transition from balanced island mode of two loaded sections to the island and unbalanced mode of the first section; transition from the balanced island mode of loaded sections to island synchronous section operation. These tests have demonstrated the flexibility of the developed system, as well as its ability to take into account user commands and adapt to different power system modes. Thus, the study results illustrate the ability to control a distributed generation power system based on generic artificial intelligence. Decision-making and adaptability to changing conditions make the proposed system a valuable tool to ensure network efficiency.

искусственный интеллектбольшая языковая модельинтеллектуальный агентсистема управленияGPTмалая распределенная генерацияSmart Gridartificial intelligencelarge language modelintelligent agentcontrol systemGPTsmall distributed generationSmart GridReferencesProstejovsky A.M., Brosinsky C., Heussen K., Westermann D., Kreusel J., Marinelli M. The future role of human operators in highly automated electric power systems // Electric Power Systems Research. 2019. Vol. 175. P. 105883. https://doi.org/10.1016/J.EPSR.2019.105883.Prostejovsky A.M., Brosinsky C., Heussen K., Westermann D., Kreusel J., Marinelli M. The future role of human operators in highly automated electric power systems. Electric Power Systems Research. 2019;175:105883. https://doi.org/10.1016/J.EPSR.2019.105883.Liu You-bo, Liu Jun-yong, Taylor G., Liu Ting-jian, Gou Jing, Zhang Xi. Situational awareness architecture for smart grids developed in accordance with dispatchers thought process: a review // Frontiers of Information Technology and Electronic Engineering. 2016. Vol. 17. Iss. 11. P. 1107–1121. https://doi.org/10.1631/FITEE.1601516.Liu You-bo, Liu Jun-yong, Taylor G., Liu Ting-jian, Gou Jing, Zhang Xi. Situational awareness architecture for smart grids developed in accordance with dispatchers thought process: a review. Frontiers of Information Technology and Electronic Engineering. 2016;17(11):1107-1121. https://doi.org/10.1631/FITEE.1601516.Uddin M., Mo Huadong, Dong Dezun, Elsawah S., Zhu Jianguo, Guerrero J.C. Microgrids: a review, outstanding issues and future trends // Energy Strategy Reviews. 2023. Vol. 49. P. 101127. https://doi.org/10.1016/j.esr.2023.101127.Uddin M., Mo Huadong, Dong Dezun, Elsawah S., Zhu Jianguo, Guerrero J.C. Microgrids: a review, outstanding issues and future trends. Energy Strategy Reviews. 2023;49:101127. https://doi.org/10.1016/j.esr.2023.101127.Liu Z., Zhang Q. Study on the development mode of national power grid of China // Proceedings of the Chinese Society of Electrical Engineering. 2013. Vol. 33. Iss. 7. P. 1–10.Liu Z., Zhang Q. Study on the development mode of national power grid of China. Proceedings of the Chinese Society of Electrical Engineering. 2013;33(7):1-10.Yu Xinghuo, Cecati C., Dillon T., Simoes M.G. The new frontier of smart grids // Industrial Electronics Magazine. 2011. Vol. 5. Iss. 3. P. 49–63. https://doi.org/10.1109/MIE.2011.942176.Yu Xinghuo, Cecati C., Dillon T., Simoes M.G. The new frontier of smart grids. Industrial Electronics Magazine. 2011;5(3):49-63. https://doi.org/10.1109/MIE.2011.942176.Yoon Deunsol, Hong Sunghoon, Lee Byung-Jun, Kim Kee-Eung. Winning the L2RPN challenge: power grid management via Semi-Markov afterstate actor-critic // International Conference on Learning Representations. 2020. P. 720–729.Yoon Deunsol, Hong Sunghoon, Lee Byung-Jun, Kim Kee-Eung. Winning the L2RPN challenge: power grid management via Semi-Markov afterstate actor-critic. In: International Conference on Learning Representations. 2020;720-729.Damjanović I., Pavić I., Puljiz M., Brcic M. Deep reinforcement learning-based approach for autonomous power flow control using only topology changes // Energies. 2022. Vol. 15. Iss. 19. P. 6920. https://doi.org/10.3390/en15196920. EDN: AYWUOU.Damjanović I., Pavić I., Puljiz M., Brcic M. Deep reinforcement learning-based approach for autonomous power flow control using only topology changes. Energies. 2022;15(19):6920. https://doi.org/10.3390/en15196920. EDN: AYWUOU.Wang G., Guo J., Ma S., Shang Y., Guo Q., Bu G., et al. Preliminary study of power system enhanced intelligence analysis // Proceedings of the Chinese Society of Electrical Engineering. 2020. Vol. 40. Iss. 16. P. 5079–5087.Wang G., Guo J., Ma S., Shang Y., Guo Q., Bu G., et al. Preliminary study of power system enhanced intelligence analysis. Proceedings of the Chinese Society of Electrical Engineering. 2020;40(16):5079-5087.He Ziyao, Song Yunpeng, Yang Gang, Qiao Wen, Wang Haoming, Wang Jinyun, et al. A dialogue-driven framework for intelligent and automated simulation of complex power systems // CSEE Journal of Power and Energy Systems. 2023. P. 1–2.He Ziyao, Song Yunpeng, Yang Gang, Qiao Wen, Wang Haoming, Wang Jinyun, et al. A dialogue-driven framework for intelligent and automated simulation of complex power systems. CSEE Journal of Power and Energy Systems. 2023;1-2.Marot A., Rozier A., Dussartre M., Crochepierre L., Donnot B. Towards an AI assistant for power grid operators // HHAI2022: Augmenting Human Intellect. IOS Press, 2022. P. 79–95.Marot A., Rozier A., Dussartre M., Crochepierre L., Donnot B. Towards an AI assistant for power grid operators. In: HHAI2022: Augmenting Human Intellect. IOS Press; 2022, p. 79-95.Gamage G., Kahawala S., Mills N., De Silva D., Manic M., Alahakoon D., et al. Augmenting industrial chatbots in energy systems using ChatGPT generative AI // LEEE 32nd International Symposium on Industrial Electronics. 2023. P. 1–3. https://doi.org/10.1109/ISIE51358.2023.10228101.Gamage G., Kahawala S., Mills N., De Silva D., Manic M., Alahakoon D., et al. Augmenting industrial chatbots in energy systems using ChatGPT generative AI. In: LEEE 32nd International Symposium on Industrial Electronics. 2023;1-3. https://doi.org/10.1109/ISIE51358.2023.10228101.Tafferner Z., Balázs I., Krammer O., Géczy A. Can ChatGPT help in electronics research and development? A case study with applied sensors // Sensors. 2023. Vol. 23. Iss. 10. P. 4879. https://doi.org/10.3390/s23104879.Tafferner Z., Balázs I., Krammer O., Géczy A. Can ChatGPT help in electronics research and development? A case study with applied sensors. Sensors. 2023;23(10):4879. https://doi.org/10.3390/s23104879.Marot A., Donnot B., Tazi S., Panciatici P. Expert system for topological remedial action discovery in smart grids. Режим доступа: https://hal.science/hal-01897931/file/_LARGE__bf_Expert_System_for_topological_remedial_action_discovery_in_smart_grids.pdf (дата обращения: 07.02.2025).Marot A., Donnot B., Tazi S., Panciatici P. Expert system for topological remedial action discovery in smart grids. Available from: https://hal.science/hal-01897931/file/_LARGE__bf_Expert_System_for_topological_remedial_action_discovery_in_smart_grids.pdf [Accessed 7th February 2025].Chen Yong, Peng Xiangjun, Xu Xianchao, Wu Hao. Deep reinforcement learning based applications in smart power systems // Journal of Physics: Conference Series. 2021. Vol. 1881. Iss. 2. P. 022051. https://doi.org/10.1088/1742-6596/1881/2/022051. EDN: ONJNXB.Chen Yong, Peng Xiangjun, Xu Xianchao, Wu Hao. Deep reinforcement learning based applications in smart power systems. Journal of Physics: Conference Series. 2021;1881(2):022051. https://doi.org/10.1088/1742-6596/1881/2/022051. EDN: ONJNXB.Li Ran, Pu Chuanqing, Tao Junyi, Li Canbing, Fan Feilong, Xiang Yue, et al. LLM-based frameworks for power engineering from routine to novel tasks // Social Science Research Network. 2023. https://doi.org/10.2139/ssrn.4741095.Li Ran, Pu Chuanqing, Tao Junyi, Li Canbing, Fan Feilong, Xiang Yue, et al. LLM-based frameworks for power engineering from routine to novel tasks. Social Science Research Network. 2023. https://doi.org/10.2139/ssrn.4741095.Nair D.R., Nair M.G., Thakur T. A smart microgrid system with artificial intelligence for power sharing and power quality improvement // Energies. 2022. Vol. 15. Iss. 15. P. 5409. https://doi.org/10.3390/en15155409.Nair D.R., Nair M.G., Thakur T. A smart microgrid system with artificial intelligence for power sharing and power quality improvement. Energies. 2022;15(15):5409. https://doi.org/10.3390/en15155409.Pribylskii V., Armeev D.V. AI control system for power grid based on distributed energy resources // IEEE 25th International Conference of Young Professionals in Electron Devices and Materials (Novosibirsk, 28 June – 2 July 2024). Novosibirsk: Novosibirsk State Technical University, 2024. Vol. 25. P. 1420–1425. https://doi.org/10.1109/EDM61683.2024.10614965. EDN: JLWEOZ.Pribylskii V., Armeev D.V. AI control system for power grid based on distributed energy resources. In: IEEE 25th International Conference of Young Professionals in Electron Devices and Materials. 28 June – 2 July 2024, Novosibirsk. Novosibirsk: Novosibirsk State Technical University; 2024, vol. 25, p. 1420-1425. https://doi.org/10.1109/EDM61683.2024.10614965. EDN: JLWEOZ.Wei Jason, Wang Xuezhi, Schuurmans D., Bosma M., Ichter B., Xia Fei, et al. Chain-of-thought prompting elicits reasoning in large language models // 36th Conference on Neural Information Processing Systems. 2022. Vol. 35. P. 24824–24837. https://doi.org/10.48550/arXiv.2201.11903.Wei Jason, Wang Xuezhi, Schuurmans D., Bosma M., Ichter B., Xia Fei, et al. Chain-of-thought prompting elicits reasoning in large language models. In: 36th Conference on Neural Information Processing Systems. 2022;35:24824-24837. https://doi.org/10.48550/arXiv.2201.11903.Zhao Xufeng, Li Mengdi, Lu Wenhao, Weber Cornelius, Lee Jae Hee, Chu Kun, et al. Enhancing zero-shot chain-of-thought reasoning in large language models through logic // Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation. 2023. https://doi.org/10.48550/arXiv.2309.13339.Zhao Xufeng, Li Mengdi, Lu Wenhao, Weber Cornelius, Lee Jae Hee, Chu Kun, et al. Enhancing zero-shot chain-of-thought reasoning in large language models through logic. In: Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation. 2023. https://doi.org/10.48550/arXiv.2309.13339.Giray L. Prompt engineering with ChatGPT: a guide for academic writers // Annals of Biomedical Engineering. 2023. Vol. 51. Iss. 12. P. 2629–2633. https://doi.org/10.1007/s10439-023-03272-4.Giray L. Prompt engineering with ChatGPT: a guide for academic writers. Annals of Biomedical Engineering. 2023;51(12):2629-2633. https://doi.org/10.1007/s10439-023-03272-4.

The authors declare that there are no conflicts of interest present.