Multivariable Generalized Predictive Control with Measurement Noise Rejection and Speed Ripple Mitigation for PMSM Drives

Generalised predictive control (GPC) is known for its good dynamic performance and long prediction horizon, but the performance can be severely deteriorated when measurement noises exist due to the wide control bandwidth of GPC. Meanwhile, the unmodelled periodic disturbances can cause ripples to the output variables. In this study, a practical permanent magnet synchronous motor (PMSM) drive system with speed measurement noise and periodic disturbances is considered, for which a novel multivariable GPC method is proposed to achieve both good dynamic performance and speed ripple mitigation. The proposed method has a simple structure, since the traditional cascaded speed and current control loops are replaced by a non-cascaded one. To reject the measurement noise without jeopardising the system stability, an internal low-pass filter is embedded in the GPC. Meanwhile, external resonant loops are added to the GPC to mitigate the low-order speed ripples caused by the periodic disturbances. Furthermore, a deadbeat-based current constraint method is proposed to avoid overcurrent during transient processes. Theoretical stability analysis of the proposed method is presented. Experimental results show that the proposed method has good steady-state and dynamic performances, including measurement noise rejection and speed ripple mitigation.