Circulating Current Suppression in Modular Multilevel Converters Based on Hybrid Model Predictive Control

The modular multilevel converter (MMC) has unique topological characteristics and has gained significant popularity in medium-voltage applications. However, during MMC operation, circulating currents inevitably arise, exacerbating arm current distortion, causing additional losses in the converter system, and thereby increasing system costs. This paper primarily addresses the circulating current issue in traditional half-bridge MMCs by introducing a control strategy combining model predictive control and proportional resonance controllers. First, a value function is established using a discrete prediction model of the output current, followed by a rolling optimization that combines a capacitor voltage sorting strategy to determine the duty cycles of each submodule in the arm. Secondly, a proportional resonance controller (PR) is designed to eliminate the second- and fourth-order harmonic components in the circulating current. Finally, the output of the resonance controller is used to correct the duty cycles, which are then compared with the PWM triangular carrier to generate more switching actions that suppress the circulating current. The effectiveness of the strategy is experimentally verified. The results show that the proposed method yields better output characteristics, smaller capacitor voltage fluctuations, and significantly suppresses harmonic components in the arm currents.