Parameter Tuning of Sliding Mode Controller Based on Reinforcement Learning of Multiple Intelligences
The method of using multi-intelligent reinforcement learning to rectify the parameters of the sliding-mode controller was proposed for the problem that it is difficult to rectify the parameters of the sliding-mode controller in the permanent magnet synchronous motor system with many parameters and a large range,which leads to poor control of the permanent magnet synchronous motor.The method was based on a shared reward for each parameter of the controller,which effectively avoided the dimensional catastrophe caused by the large difference in the range of different parameters selected by the intelligent algorithm.The results show that the multi-arm slot machine algorithm with multiple intelligences has obvious advantages over the genetic algorithm in terms of overshoot,response speed,anti-interference ability and robustness in tuning the speed sliding mode controller.It is verified that this method can effectively solve the problem that the parameters of sliding mode controller are difficult to be adjusted.
万方数据
维普
