New Predictive Current Control for Modular Multilevel Converters with Revised Prediction Model Considering Common-Mode Voltage

The modular multilevel converter (MMC) is an attractive topology for high power applications. Model predictive control (MPC) becomes a very promising alternative for MMCs. The conventional MPC solution for MMC usually uses separate control for each phase so to reduce the calculation burden. However, such technique cannot take the common-mode voltage impacts into consideration, resulting in an inaccurate prediction and hence poor control precision. To tackle these problems, this work proposes two novel predictive current control (PCC) methods, i.e., Jacobi-irritation-based PCC (J-PCC), and steepest-descent PCC (SD-PCC). Both methods consider common-mode voltage impacts in system modeling, and have reduced computational burden. Additionally, an active submodule capacitor average voltage control technique is realized based on the aforementioned methods, achieving good submodule voltage balance of all the arms in the MMC. Simulation results, at different test scenarios, confirm the effectiveness of the proposal.