Improved model predictive control for modular multilevel converter with reduced voltage sensors based on Kalman filter

Modular multilevel converters (MMCs) are used in medium and high power applications due to various advantages, such as modularity, scalability, high efficiency, and low THD. In this paper, an improved model predictive control (IMPC) is used to control the converter output current and the circulating current. Using a discrete mathematical model of MMC and the neighboring index values with respect to their previously applied values, the calculation burden can be reduced rapidly even the number of sub-modules (SMs) increases. By applying this method, the control system complexity and the number of switching states are reduced. In addition, a Kalman filter (KF) is applied to estimate the capacitor voltage of the SMs, by using two arm voltage sensors in each leg of MMC. The combination of IMPC and KF reduces the computational burden and the number of voltage sensors. The efficiency and superiority of the proposed method are verified by simulation results of 5- and 11-level three-phase MMC and practical implementation of a laboratory single-phase converter with four SMs in each arm.