A Modulated Model Predictive Control Method Based on Vector Analysis for Four-State Multilevel Converters

The paper presents a new multi-objective modulated model predictive control (MMPC) method for power converters with four different model states. The proposed method is based on a new concept of composing an error vector with fundamental state vectors. It can significantly minimize voltage and current ripples while retaining the nonlinear and dynamic performance of conventional MPC methods. Simulation and experimental results on a SiC-based 3-level flying capacitor(FC) converter are provided to validate this concept, and indicate a much reduced control error with the new method.