期刊
IEEE TRANSACTIONS ON POWER SYSTEMS
卷 36, 期 6, 页码 5143-5152出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TPWRS.2021.3070161
关键词
Generators; Heuristic algorithms; Power system dynamics; Cost function; Wind power generation; Upper bound; Research and development; Distributed reinforcement learning; economic dispatch; multi-agent system; singularly perturbed system
资金
- National Natural Science Foundation of China [61773260, 61590925]
- Key R&D Program of the Ministry of Science and Technology [2018YFB1305902]
The study presents a unified distributed RL solution for both static and dynamic economic dispatch problems, using a projection method to generate feasible actual actions. Distributed Hysteretic Q-learning coordinates agents effectively, while the algorithm handles continuous action space and power loads without using function approximations. Theoretical analysis and comparative simulation studies demonstrate the algorithm's convergence and optimality.
A unified distributed reinforcement learning (RL) solution is offered for both static and dynamic economic dispatch problems (EDPs). Each agent is assigned with a fixed, discrete, virtual action set, and a projection method generates the feasible, actual actions to satisfy the constraints. A distributed algorithm, based on singularly perturbed system, solves the projection problem. A distributed form of Hysteretic Q-learning achieves coordination among agents. Therein, the Q-values are developed based on the virtual actions, while the rewards are produced by the projected actual actions. The proposed algorithm deals with continuous action space and power loads without using function approximations. Theoretical analysis and comparative simulation studies verify algorithm's convergence and optimality.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据