Adaptive Neural-Network Optimal Tracking Control for Permanent-Magnet Synchronous Motor Drive System via Adaptive Dynamic Programming

Abstract

This paper presents an adaptive neural-network optimal tracking control (ANOTC) scheme for permanent-magnet synchronous motor (PMSM) servo drive with uncertain dynamics via adaptive dynamic programming (ADP). The proposed ANOTC scheme consists of an adaptive steady-state controller, an adaptive optimal feedback controller and a robust controller. The adaptive steady-state controller is designed for attaining the targeted tracking response during the steady-state. The adaptive optimal feedback controller is designed for stabilizing the dynamics of tracking error at the transient in an optimal manner. Accordingly, critic and actor neural-networks are employed for facilitating the online solution of the Hamilton-Jacobi-Bellman (HJB) equation for approximating the adaptive optimal control laws via ADP method. Further, the robust controller is developed for compensating the approximation errors of neural-network (NN). Based on Lyapunov approach, the closed-loop stability of the PMSM servo drive system is proved to demonstrate that the proposed ANOTC scheme can ensure the system state tracking the targeted trajectory effectively. The proposed ANOTC scheme validation is performed via experimental analysis. From the experimental validation results, the PMSM servo drive dynamic behavior using the proposed ANOTC scheme can attain the optimal control response regardless the compounded disturbances and parameter uncertainties.