Zheng Gong, Chengquan Yang, Yifeng Su, Changming Zheng
Model predictive control (MPC) has been widely adopted in modular multilevel converters (MMC) due to its multi-objective capabilities and rapid dynamic response. However, it faces challenges such as significant computational burden, difficulty in adjusting weighting factors, and variable switching frequency. Additionally, the presence of numerous redundant switching states further complicates the application of space vector modulation in MMC. To address these issues, this paper proposes an optimal switching sequence MPC (OSS-MPC) for MMC. First, a method for determining the candidate switching sequence is developed to identify specific switching states while minimizing computational complexity. Next, a predictive model based on the gradient of output current is introduced to reduce current ripple and improve tracking accuracy. Finally, circulating current suppression is achieved by integrating the predictive model of arm unbalanced voltage with the OSS. The proposed OSS-MPC exhibits excellent steady-state performance and a reduced computational burden, while preserving the rapid response characteristic of MPC. Simulation and experimental results from a prototype with 24 submodules demonstrate the feasibility of the proposed method.