基于强化学习的低速转动PMSM高性能伺服控制  

High-performance servo control of PMSM under low-speed rotation conditions based on reinforcement learning

作　　者：李雅楠 杨森 李晓峰 Ll Yanan;YANG Sen;LI Xiaofeng(School of Astronaufics and Aeronautics,Univ ersity of Electronic Science and'Technology of Chin a,Chengdu611731,China)

机构地区：[1]电子科技大学航空航天学院,成都611731

出　　处：《电子科技大学学报》2025年第2期266-273,共8页Journal of University of Electronic Science and Technology of China

摘　　要：实现对永磁同步电机(PMSM)在低速转动工况下的高精度伺服控制是建立高性能星载激光通信链路的前提,其技术关键是对永磁同步电机转动状态的非线性特性进行精确描述。该文设计了一种具有能够处理连续动作空间特点的DDPG非线性控制器,采用梯度下降法分别训练评价神经网络和动作神经网络,实现了对非线性映射的精确拟合。Simulink仿真结果表明:和传统的比例-积分线性控制器相比较,DDPG控制器在跟踪参考低速信号时响应时间和稳定时间更短、跟踪误差更小;在施加扭矩时q轴电流响应更快,d轴电流波动更小,低速工况条件下的PMSM伺服控制性能得到了有效提高。Achieving high-precision servo control performance of permanent magnet synchronous motors(PMSM)under low-speed rotation conditions is a prerequisite for establishing high-performance satellite lasercommunication links.The key technology lies in the accurate description of the nonlinear characteristics of thePMSM rotational state.This paper designs a deep deterministic policy gradient(DDPG)nonlinear controllercapable of handling continuous action spaces,utilizing gradient descent methods to train both the critic and actorneural networks separately,thereby achieving precise fitting of nonlinear mappings.Through Simulink simulations,the results demonstrate that compared to traditional proportional-integral linear controllers,the DDPG controlexhibits shorter response and settling times,smaller tracking errors when following reference low-speed signals;italso shows faster q-axis current response and smaller d-axis current fluctuations when torque is applied,effectivelyimproving the PMSM servo control performance under low-speed operating conditions.

关 键 词：永磁同步电机 低速转动 深度确定性策略梯度 非线性 伺服控制性能 

分 类 号：TM3[电气工程—电机]