关键词: 永磁同步电机, 脉冲神经网络, 强化学习, 双延迟确定性梯度算法

Abstract: To enhance the control performance of permanent magnet synchronous motor ( PMSM ) under multiple operating conditions and address the limitations of conventional PI controllers, such as fixed parameters and poor adaptability,this paper innovatively proposes an intelligent control method ( SNN-TD3 ) that integrates spiking neural networks ( SNN) with the twin delayed deep deterministic policy gradient algorithm ( TD3) . In this method,the SNN serves as the Actor network,replacing the PI controllers in the speed loop and q-axis current loop of traditional vector control. By leveraging the event-driven characteristics and temporal encoding capabilities of SNNs, the dynamic response and noise robustness of the system are significantly improved. Through co-simulation in MATLAB / Simulink and Python, simulation experiments are conducted under various operating conditions, including no-load startup, load disturbance, sinusoidal tracking,and composite step changes. The results demonstrate that the proposed SNN-TD3 method outperforms traditional PI and standard TD3 control in terms of zero overshoot,fast response,and tracking accuracy,exhibiting superior comprehensive control performance and strong adaptability to diverse operating conditions.

Key words: permanent magnet synchronous motor,spiking neural network, reinforcement learning, twin delayed deep deterministic policy gradient algorithm