Ultra-Local Model Predictive Current Control of Permanent Magnet Synchronous Motor With Dual-Vector Based on Data-Driven Neural Networks

Chendong Zhao, Dan Wang, Zhouhua Peng, Wenjie Wu

Abstract: The traditional model predictive current control (MPCC) of permanent magnet synchronous motor (PMSM) has the advantages of simple control structure and excellent dynamic performance. However, the performance of the MPCC is significantly impacted by changes in motor parameters. The inside and outside unknown disturbance of the motor cause the parameters mismatch, which negatively affects the performance of the MPCC controller. To eliminate the effects of the parameters mismatch, the original parameter-based predictive model is replaced by an ultra-local model in this paper. An estimator based on data-driven neural network is designed to quickly and accurately estimate the total perturbation and control gain of the established ultra-local model. The proposed design solely utilizes the input and output information of the controlled system instead of relying on motor parameters, thus avoiding the negative effects of model parameters mismatch. In addition, the dual-vector mechanism and delay compensation are added to improve the control performance. Finally, the stability analysis is given, and simulated results show the availability of the proposed method.

Submission Number: 92