Paramater identification of PMSM using deep reinforecrmrnt learning

In this paper, a self-adaptive permanent magnet synchronous machines (PMSM) parameter calibration algorithm using deep reinforcement learning (DRL) is proposed, which transforms the parameter calibration problem into a Markov decision process (MDP). The performance of the proposed schemes is tested on a nonlinear PMSM model under variable loads and Gaussian noise. The results show that the proposed algorithm has better accuracy and robustness than traditional optimization algorithm.