Design and tests of reinforcement-learning-based optimal power flow solution generator

Optimal power flow (OPF) is a very traditional problem in the research field of power systems. In this paper, an OPF solution generator based on reinforcement learning (RL) is proposed. The solution process of OPF is modeled as a one-step Markov Decision Process (MDP) and is solved using the Twin De...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Hongyue Zhen, Hefeng Zhai, Weizhe Ma, Ligang Zhao, Yixuan Weng, Yuan Xu, Jun Shi, Xiaofeng He
Formato:	article
Lenguaje:	EN
Publicado:	Elsevier 2022
Materias:	Optimal power flow Reinforcement learning TD3 Parallel computing Electrical engineering. Electronics. Nuclear engineering TK1-9971
Acceso en línea:	https://doaj.org/article/c86980c64de74a089d63e76c876a750a
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:c86980c64de74a089d63e76c876a750a
record_format	dspace
spelling	oai:doaj.org-article:c86980c64de74a089d63e76c876a750a2021-12-04T04:35:06ZDesign and tests of reinforcement-learning-based optimal power flow solution generator2352-484710.1016/j.egyr.2021.11.126https://doaj.org/article/c86980c64de74a089d63e76c876a750a2022-04-01T00:00:00Zhttp://www.sciencedirect.com/science/article/pii/S2352484721012737https://doaj.org/toc/2352-4847Optimal power flow (OPF) is a very traditional problem in the research field of power systems. In this paper, an OPF solution generator based on reinforcement learning (RL) is proposed. The solution process of OPF is modeled as a one-step Markov Decision Process (MDP) and is solved using the Twin Delayed Deep Deterministic policy gradient (TD3) algorithm. A warm-up training mechanism is adopted to realize better initialization of neural networks. Parallel computing is utilized to expand the searching range and improve training efficiency. Numerical tests are carried out in the IEEE-39 system. The results prove the correctness and efficiency of the proposed algorithm. The actor (policy) network of the well-trained agent can serve as a fast optimal power flow solution generator and can be applied to online scenarios.Hongyue ZhenHefeng ZhaiWeizhe MaLigang ZhaoYixuan WengYuan XuJun ShiXiaofeng HeElsevierarticleOptimal power flowReinforcement learningTD3Parallel computingElectrical engineering. Electronics. Nuclear engineeringTK1-9971ENEnergy Reports, Vol 8, Iss , Pp 43-50 (2022)
institution	DOAJ
collection	DOAJ
language	EN
topic	Optimal power flow Reinforcement learning TD3 Parallel computing Electrical engineering. Electronics. Nuclear engineering TK1-9971
spellingShingle	Optimal power flow Reinforcement learning TD3 Parallel computing Electrical engineering. Electronics. Nuclear engineering TK1-9971 Hongyue Zhen Hefeng Zhai Weizhe Ma Ligang Zhao Yixuan Weng Yuan Xu Jun Shi Xiaofeng He Design and tests of reinforcement-learning-based optimal power flow solution generator
description	Optimal power flow (OPF) is a very traditional problem in the research field of power systems. In this paper, an OPF solution generator based on reinforcement learning (RL) is proposed. The solution process of OPF is modeled as a one-step Markov Decision Process (MDP) and is solved using the Twin Delayed Deep Deterministic policy gradient (TD3) algorithm. A warm-up training mechanism is adopted to realize better initialization of neural networks. Parallel computing is utilized to expand the searching range and improve training efficiency. Numerical tests are carried out in the IEEE-39 system. The results prove the correctness and efficiency of the proposed algorithm. The actor (policy) network of the well-trained agent can serve as a fast optimal power flow solution generator and can be applied to online scenarios.
format	article
author	Hongyue Zhen Hefeng Zhai Weizhe Ma Ligang Zhao Yixuan Weng Yuan Xu Jun Shi Xiaofeng He
author_facet	Hongyue Zhen Hefeng Zhai Weizhe Ma Ligang Zhao Yixuan Weng Yuan Xu Jun Shi Xiaofeng He
author_sort	Hongyue Zhen
title	Design and tests of reinforcement-learning-based optimal power flow solution generator
title_short	Design and tests of reinforcement-learning-based optimal power flow solution generator
title_full	Design and tests of reinforcement-learning-based optimal power flow solution generator
title_fullStr	Design and tests of reinforcement-learning-based optimal power flow solution generator
title_full_unstemmed	Design and tests of reinforcement-learning-based optimal power flow solution generator
title_sort	design and tests of reinforcement-learning-based optimal power flow solution generator
publisher	Elsevier
publishDate	2022
url	https://doaj.org/article/c86980c64de74a089d63e76c876a750a
work_keys_str_mv	AT hongyuezhen designandtestsofreinforcementlearningbasedoptimalpowerflowsolutiongenerator AT hefengzhai designandtestsofreinforcementlearningbasedoptimalpowerflowsolutiongenerator AT weizhema designandtestsofreinforcementlearningbasedoptimalpowerflowsolutiongenerator AT ligangzhao designandtestsofreinforcementlearningbasedoptimalpowerflowsolutiongenerator AT yixuanweng designandtestsofreinforcementlearningbasedoptimalpowerflowsolutiongenerator AT yuanxu designandtestsofreinforcementlearningbasedoptimalpowerflowsolutiongenerator AT junshi designandtestsofreinforcementlearningbasedoptimalpowerflowsolutiongenerator AT xiaofenghe designandtestsofreinforcementlearningbasedoptimalpowerflowsolutiongenerator
_version_	1718372995350659072

Design and tests of reinforcement-learning-based optimal power flow solution generator

Ejemplares similares