深度强化学习驱动的双馈抽蓄抽水工况下调频控制  

Frequency Regulation of Doubly-fed Induction Machine Pumped Storage Hydro in Pumping Mode Driven by Deep Reinforcement Learning

作　　者：劳文洁 史林军[1] 李杨 吴峰[1] 林克曼 LAO Wenjie;SHI Linjun;LI Yang;WU Feng;LIN Keman(College of Energy and Electrical Engineering,Hohai University,Nanjing 210098,China)

机构地区：[1]河海大学能源与电气学院,南京210098

出　　处：《电力系统及其自动化学报》2023年第12期59-70,共12页Proceedings of the CSU-EPSA

基　　金：国家自然科学基金中英合作资助项目(52061635102)。

摘　　要：为改善新型电力系统的频率特性,利用抽水工况下双馈抽水蓄能机组功率可调的特点,提出基于多智能体深度确定性策略梯度算法的系统频率控制方法。首先,基于抽水工况下双馈抽水蓄能的频率控制策略,构建多能互补系统的频率控制模型;其次,以提高各区域控制性能标准指标为目标,利用改进的多智能体深度确定性策略梯度算法优化各机组的自动发电控制指令。算例分析表明,抽水工况下双馈抽水蓄能参与调频可显著改善系统的频率特性,且所提频率控制方法的鲁棒性和可靠性优于传统控制。To improve the frequency characteristics of novel power systems,a system frequency control method based on the multi-agent deep deterministic policy gradient(MADDPG)algorithm is proposed by utilizing the characteristics of a doubly-fed induction machine pumped storage hydro(DFIM-PSH)unit,whose power can be adjusted in pumping mode.First,based on the frequency control strategy for DFIM-PSH in pumping mode,a frequency control model of the hybrid energy system is constructed.Second,to improve the control performance standard indicators of each region,the improved MADDPG algorithm is used to optimize the AGC control instructions of each unit.The case studies show that the participation of DFIM-PSH in frequency regulation in pumping mode can significantly improve the system’s frequency characteristics,and the proposed frequency control method is better than traditional control methods in terms of robustness and reliability.

关 键 词：调频 双馈抽水蓄能机组 多智能体深度确定性策略梯度算法 多能互补系统 控制性能标准 

分 类 号：TM614[电气工程—电力系统及自动化]