Two-Phase Model Predictive Direct Duty-Cycle Control for PMSM With Off-Line Lookup Table

Abstract

Finite control set model predictive control has poor steady-state performance due to limited voltage vectors. To improve the steady-state performance, multiple vectors are used to synthesize more accurate voltage vectors, but the selection of optimal vectors is always complex. In this paper, a two-phase model predictive direct duty-cycle control based on off-line lookup table (OLT) is constructed. The proposed prediction model converts three-phase duty cycles to two-phase duty cycles, which reduces the number of duty cycles calculations. In addition, to optimize the reference voltage vectors space, a new flux linkage deviation mechanism is proposed where the virtual reference vectors can be determined with low complexity by using OLT. Then, due to further improve the steady-state accuracy and ensure low complexity, a new flux linkage constraint is proposed, which restricts the flux linkage while reducing the torque ripple. More importantly, no weight factor is needed to balance the torque and stator flux. Finally, the process of the multi-stage and multi-duty-cycle calculation is converted into the process of a one-stage and small range single-variable enumeration selection, which significantly reduces calculation cost while improving the steady-state performance. The feasibility and effectiveness ofthe proposed method isverified bytheoryand experiments.