Dual-Mode Model Predictive Control for Neutral-Point-Clamped Inverter-Fed PMSM With Dynamic Region Partition

Model predictive control for the neutral-point-clamped (NPC) inverter generally cannot optimize the transient neutral-point potential (NPP) balancing and the steady-state performance at the same time. Besides, neglecting the vector shift caused by the real-time NPP floating seriously affects the control performance. To address these issues, this article proposes a dual-mode model predictive control (DM-MPC) method to comprehensively improve system performance by freely switching the control strategies depending on the transient or steady-state mode. When entering the transient mode, the NPP balancing is accelerated by eliminating weight factors and narrowing the candidate vector selection. For the steady-state mode of the proposed technique, virtual vectors are precisely constructed by real-time NPP feedback, while the optimal vector combination and its accurate duration are determined according to the triangle area rule. These ensure superior current quality for balanced and unbalanced NPP conditions. Finally, the given experimental results on an NPC inverter-fed motor drive system have proven the feasibility and advantages of the proposed DM-MPC method.