Abstract

Model predictive control (MPC) has been widely investigated in modular multilevel converters (MMCs) due to its superiority in achieving multiple control objectives. The three-phase model-based MPC, which contains the common-mode voltage in the output current dynamic model and considers interaction among phases, shows better performance than the conventional per-phase model-based predictive control in a three-phase MMC system. However, it suffers from a heavy computational burden as the number of submodules (SMs) increases. To address this issue, this article first analyzes the relationship among the numbers of inserted SMs, the controllability of dc-link current, and circulating currents. Then, according to this analysis, two simplified MPC methods based on the three-phase model with reduced computational burden are proposed. Specifically, fewer insertion index combinations are selected in advance to ensure good output currents, controllable dc-link, and circulating currents. The effectiveness of the proposed methods is verified through experimental results.