A Model Predictive Control Strategy with Virtual Vectors Visualization Analysis for NPC Converter Without Weighting Factor

To address the high current ripple observed in traditional model predictive control (MPC) of neutral-pointclamped (NPC) converters. A visualization analysis method is proposed to evaluate the voltage vector diagram and the errors of individual basic voltage vectors, with a virtual vector introduced at the position of maximum error. However, the MPC for NPC also suffers from neutral point (NP) voltage imbalance. The basic and virtual vectors (VVs) are partitioned into two candidate groups, from which a subset is selected based on NP voltage balance considerations. Iterative selection among candidate voltage vectors is performed, eliminating the need for designing weight factors in the conventional cost function. Moreover, to mitigate the increased computational complexity due to the additional VVs, the voltage vector space is divided into six central sectors and six peripheral sectors, within which the optimal voltage vector is iteratively determined. Simulation results validate the effectiveness of the proposed method.