Closed-Loop Predictively Optimizing Control for Modular Multilevel Converter with Parallel Connectivity

This paper adopts the bilinear modeling method to design a coherent model for modular multilevel converters with parallel connectivity and proposes an optimized closed-loop control combining proportional resonant (PR) control, nearest level modulation, and online-optimization scheduling. A bilinear mathematical model of the MMSPC is developed to predict the behavior of MMSPC one step ahead and thus reduce the computation burden of the proposed closed-loop control. PR control guarantees small steady state error and rapid dynamic response. Thanks to PR control, simple nearest level modulation can replace digital pulse-width modulation or sigma-delta modulation and decrease switching loss sharply. Moreover, a novel online-optimization scheduler replaces previously used offline optimization to improve module balance. The effectiveness of the proposed closed-loop control is verified by an accurate circuit simulation setup that represents the experimental prototype.