Parameter Tuning of PMSM Sliding Mode Control Based on Multi-agent Reinforcement Learning
摘要
Firstly, this paper establishes a multi-agent reinforcement learning parameter tuning framework for PMSM sliding mode controller, and uses TD3 as the algorithm of agent learning. Secondly, the final parameters are obtained by training the model, which reasonably setting the observation value and reward function of reinforcement learning. Finally, the effectiveness of parameter tuning using multi-agent reinforcement learning is verified by simulation and experiment.