Nonlinear Model Predictive Speed Control with Variable Predictive Horizon for PMSM Rotor Position

Abstract

The predictive horizon in the model predictive control (MPC) or nonlinear MPC (NMPC) is a constant value which limits the performance improvement of the system. A nonlinear model predictive speed control (NMPSC) with variable predictive horizon during each sampling period is proposed in this paper, and applied into the rotor position control system of permanent magnet synchronous motor (PMSM). The predictive horizon can be decreased by the proposed method if the system does not need that long predictive horizon necessarily. The proposed method is compared with the improved NMPSC strategy with extra weighting factors, and leads to improvements in the tracking process, as well as in the robustness with stator resistance and magnet flux linkage mismatches. Even with rated load torque, the proposed method still exhibits in high servo stiffness. Simulation and experimental results indicate that the tracking performance with sinusoidal reference which makes system continuous operating in transient state is feasible and effective.