微网储能侧DC-DC变换器的强化学习自抗扰控制策略  

REINFORCEMENT LEARNING ACTIVE DISTURBANCE REJECTION CONTROL STRATEGY FOR MICROGRID ENERGY STORAGE SIDE DC-DC CONVERTER

作　　者：马幼捷[1] 刘熠铭 周雪松[1] 王博[1] 陶珑 问虎龙 Ma Youjie;Liu Yiming;Zhou Xuesong;Wang Bo;Tao Long;Wen Hulong(Tianjin Key Laboratory of New Energy Power Conversion,Transmission and Intelligent Control,Tianjin University of Technology,Tianjin 300384,China;Tianjin Ruineng Electric Co.,Ltd.,Tianjin 300381,China)

机构地区：[1]天津理工大学天津市新能源电力变换传输与智能控制重点实验室,天津300384 [2]天津瑞能电气有限公司,天津300381

出　　处：《太阳能学报》2025年第3期63-72,共10页Acta Energiae Solaris Sinica

基　　金：国家自然科学重点基金(U23B20142);天津市研究生科研创新实践项目(2022BKYZ036)。

摘　　要：直流微电网电压稳定性研究是新型电力系统面临的关键问题。针对微电网系统中直流母线电压波动大和抗干扰能力弱等缺陷,该文提出一种由Q-learning算法赋能的DC-DC变换器自抗扰控制策略。通过引入线性扩张状态观测器,实现对模型内部摄动与外部扰动量的精确估计与补偿,利用Q-learning算法实现控制策略参数自适应优化,从而更高效地维持输出电压稳定。基于理论分析,推导范数意义下的Q-learning算法收敛性,并运用Lyapunov理论判据证明线性自抗扰的稳定性。最后,通过仿真对比该文提出的控制策略、线性自抗扰控制与双闭环PI控制在不同工况下的结果,充分验证该策略在提升DC-DC变换器抗扰能力和鲁棒水平的高效性与优越性。The study of voltage stability in DC microgrids is a key issue faced by new power systems.This paper proposes a active disturbance rejection control strategy for DC-DC converters empowered by Q-learning algorithm to address the shortcomings of large DC bus voltage fluctuations and weak anti-interference ability in microgrid systems.By introducing a linear expansion state observer,precise estimation and compensation of internal and external disturbances in the model are achieved.Q-learning algorithm is used to achieve adaptive optimization of control strategy parameters,thereby maintaining output voltage stability more efficiently.Based on theoretical analysis,the convergence of Q-learning algorithm in the norm sense was derived,and the stability of linear active disturbance rejection was proved using Lyapunov theory criterion.Finally,by comparing the results of the proposed control strategy,linear active disturbance rejection control,and dual closed-loop PI control under different operating conditions through simulation,the efficiency and superiority of this strategy in improving the disturbance rejection ability and robustness level of DC-DC converters are fully verified.

关 键 词：微电网 DC-DC变换器 自抗扰控制 Q-learning算法 Lyapunov判据 

分 类 号：TM615[电气工程—电力系统及自动化]