Deadbeat Model Predictive Control for Modular Multilevel Matrix Converter Under Unbalanced Grid Conditions

The modular multilevel matrix converter (M3C) is a power converter used to directly connect two three-phase ac systems, with the advantages of modularity, scalability, and high redundancy. Conventional finite control set model predictive control (FCS-MPC) exhibits fast dynamic responses for the control of the M3C. However, its irregular pulse patterns lead to an unfixed switching frequency and larger ripples. In this article, a deadbeat model predictive control (DB-MPC) algorithm for the M3C is proposed to achieve a fixed switching frequency, lower current harmonic content, and lower computational burden, with the rapidity of the FCS-MPC preserved. Based on the proposed DB-MPC, a control strategy for the M3C under unbalanced grid conditions is proposed. The strategy aims to suppress the grid current distortions by maintaining constant grid active power, thus ensuring the power supply quality. Finally, the feasibility and effectiveness of the proposed strategy is verified by the experimental results obtained from a downscaled prototype consisting of 27 submodules.