摘要
大规模可再生能源和柔性负荷的接入会给分布式多区域互联电网带来强随机扰动,传统的控制方法无法有效提高由于强随机扰动所导致的电网愈来愈差的控制性能。为此,该文从自动发电控制角度提出一种面向分布式多区域互联电网的多智能体协同控制算法,即权重双Q-时延更新算法。所提算法可通过权重双Q算法来解决传统强化学习中动作探索值高估或低估的问题,并引入时延更新策略进一步提高其更新效率,进而提高其收敛性能。对改进的IEEE标准两区域负荷频率控制模型和融入大规模可再生能源的四区域互联电网模型进行仿真,仿真结果表明,所提算法能够有效提高电网的控制性能,实现分布式多区域互联电网间的协同控制,而且与传统方法相比,具有更优控制性能和更快收敛速度。
The large-scale access of renewable energy and flexible load brings strongly random disturbance to the distributed multi-area interconnected power grid.And the traditional control methods cannot improve the increasingly poor control performance of power grid caused by the strongly random disturbance.Therefore,from the perspective of automatic generation control,this paper proposed a multi-agent cooperative control algorithm for distributed multi-area interconnected power grid,namely weighted double Q-delayed update algorithm.The proposed algorithm could solve the problem of overestimation or underestimation of action exploration value in traditional reinforcement learning through weighted thought.Besides,the delayed update strategy was introduced to further improve the updating efficiency,and then the convergence performance was optimized.Through the simulations of the improved IEEE standard two-area load frequency control model and the four-area interconnected grid model incorporating large-scale renewable energy,the results showed that the proposed algorithm effectively improved the control performance of the grid,realized the cooperative control of distributed multi-area interconnected grid,and had better control performance and faster convergence speed compared with the traditional methods.
作者
李彦营
席磊
郭宜果
王昱昊
孙梦梦
金澄心
LI Yanying;XI Lei;GUO Yiguo;WANG Yuhao;SUN Mengmeng;JIN Chengxin(College of Electrical Engineering and New Energy,China Three Gorges University,Yichang 443002,Hubei Province,China;State Grid Shandong Economic&Technology Research Institute,Jinan 250000,Shandong Province,China)
出处
《中国电机工程学报》
EI
CSCD
北大核心
2022年第15期5459-5470,共12页
Proceedings of the CSEE
基金
国家自然科学基金项目(51707102)。
关键词
权重双Q
时延更新
强化学习
自动发电控制
多智能体
weighted double Q
delayed update
reinforcement learning
automatic generation control
multi-agent
