Computationally Efficient Optimization Method for Model Predictive Pulse Pattern Control of Modular Multilevel Converters

This paper proposes a computationally efficient model predictive pulse pattern control (MP3C) for Modular multilevel converter with three-phase RL-load. The MP3C combines the optimal steady-state performance of optimized pulse patterns with the fast dynamic response of model predictive control (MPC) and is a promising control method for high power converters and industrial drives. However, the high computational burden has been an obstacle to hinder its further application. In this paper, an infeasible active set method is proposed to solve the constrained quadratic programming problem for MP3C efficiently. Besides, a per-phase arm-balancing control and the sorting algorithm are applied to balance the capacitor voltage on each submodule. The simulation results show that MP3C has excellent steady-state behavior as well as the fast response during transients.