A Deadbeat Model Predictive Control of PMSM Drives for Eliminating Common Mode Voltage Spikes and Minimizing Switching Frequency

Deadbeat- based model predictive control (DB-MPC) with traditional space vector modulation (SVM) strategy is commonly applied in motor drives. Traditional SVM utilizes zero vectors, which generate large common-mode voltage (CMV). To reduce the amplitude of CMV, various modified SVM strategies that do not utilize zero vectors have been put forward in literature, and some of them can further eliminate CMV spikes caused by dead-time effect. However, with these modified SVM strategies, the switching frequency of the power devices is relatively high, which leads to large power losses. In this paper, a DB-MPC method with an enhanced SVM is proposed. A vector sequencing rule is established by analyzing the vector switching, the internal area of the vector hexagon is dynamically divided, and the optimal vector sequence with the least number of switching actions and no CMV spikes is selected. The advantage of the proposed method is that the switching frequency of the inverter is minimized on the premise of eliminating the CMV spikes. Experiments verified the feasibility and effectiveness of the proposed method.