ADP-Based Optimal Control for Systems With Mismatched Disturbances: A PMSM Application
Zhongxin Fan, Shihua Li, Rongjie Liu
Abstract
In this brief, in order to optimize the tracking accuracy and enhance the robustness of the tracking process for permanent magnet synchronous motor (PMSM), an adaptive dynamic programming (ADP)-based optimal control strategy combining with anti-disturbance control method is proposed. We consider the unknown, mismatched and time-varying load torque as a disturbance and a disturbance observer is then designed. Based on the output of the disturbance observer, a feedforward control can compensate the disturbance in the PMSM in real time. For the easy implementation of optimal control, this brief adopts the actor-critic neural network to approximate the value function and the optimal controller, respectively. A composite controller is then proposed to ensure that the speed can track the reference signal in an optimal and robust way. Finally, an experiment based on a motor towed platform is given to prove the optimality and robustness.