A Hybrid Model Predictive Control for a Seven-Level Hybrid Multilevel Converter With Independent Low-Frequency and High-Frequency Stages

Abstract

This article puts forward a hybrid model predictive control (H-MPC) that enables the decoupling of low- and high-frequency stages in a seven-level hybrid multilevel converter (7L-HMC). It first uses the sign pattern of the reference voltage vector to determine switching states of the low-frequency stage. Next, the reference vector is shifted to the internal hexagon of the original space vector diagram, which is thereby converted to a 120° oblique coordinate system, where the adjacent vectors can be rapidly selected. Then, the optimal current tracking is primarily safeguarded by the duty cycle optimization of the three adjacent vectors. Finally, an optimized symmetric switching sequence minimizing the dc voltage deviation over one control iteration is selected through the assessment of possible switching sequences that are subject to the chosen vectors with optimal duty cycles. The proposed H-MPC method can reduce both the current ripple and computational burden while achieving a constant switching frequency. Comprehensive simulation and experimental comparisons on an all-silicon-carbide 7L-HMC prototype verify the effectiveness of the proposed H-MPC.