Modular neural dynamic surface control for position tracking of permanent magnet synchronous motor subject to unknown uncertainties

Abstract

This paper is concerned with the adaptive position tracking of permanent magnet synchronous motors (PMSMs) subject to model uncertainties and unknown loads. A modular neural dynamic surface control (MNDSC) method is used to devise the position tracking controller. Specifically, a predictor module based on neural networks (NNs) is designed, which is able to fast identify the unknown nonlinearities and uncertainties of PMSMs without introducing high-frequency oscillations in the learning process. Next, a position tracking controller module is designed based on a modified dynamic surface control where a second-order nonlinear tracking differentiator (NLTD) instead of a first-order filter is used to extract the time derivatives of virtual control law. The salient features of the proposed position tracking controller for PMSMs are as follows. First, the transient performance can be improved compared with the previous control method. Second, the unknown nonlinearities of the PMSM can be approximated by the NNs. Third, the predictor module and the controller module are decoupled by using the modular design. The stability of the position tracking system cascaded by the predictor module and the controller module is proved by cascade theory and input-to-state stability theory. Finally, the performance of the proposed MNDSC strategy for PMSMs is verified through simulations.