A novel model predictive control strategy of modular multilevel converters using voltage prediction

The model predictive control (MPC) is promising for control of the modular multilevel converter (MMC) owing to its advantages in inclusion of nonlinearities and constraints for complex systems. Existing MPC methods need to define cost functions, and the calculation burden will increase as the level of the MMC rises. This paper proposes a novel voltage predictive based MPC strategy, which combines the ac-side current and circulating current control by using forward differential voltage prediction and nearest level approximation. The chosen of switching states is intuitive according to the voltage predictive results, and there is no need to define cost functions. A pre-scaled voltage sorting method is used to balance the capacitor voltages and reduce the switching frequency. In addition, circulating current hysteresis control is introduced to further reduce the switching frequency. The performance of the proposed method is evaluated based on the simulation results of an MMC with 20 submodules per arm.