Abstract

This paper proposes a finite-set current control strategy based on deep Q-learning for permanent magnet synchronous machine (PMSM) drives. Here, the model-based current prediction of conventional model predictive control is abandoned. Instead, the proposed method selects an optimal switching action in each control period for PMSM drives by training a Deep Q-Network (DQN) to approximate the optimal Q function. Simulations are conducted to demonstrate the effectiveness of the proposed method, showing close performance compared to the conventional finite control set model predictive current control (FCS-MPCC) method.