Predictive Control of Modular Multilevel Converters: Adaptive Hybrid Framework for Circulating Current and Capacitor Voltage Fluctuation Suppression

Abstract

Modular multilevel converters (MMCs) are widely used in voltage-sourced, converter-based high-voltage DC systems due to their modular design, scalability, and fault tolerance capabilities. In MMCs, multi-variable control objectives can be employed by using model predictive control (MPC) due to its fast dynamic response and ease of implementation. Nonetheless, conventional MPC techniques for MMCs have shortcomings, including high computational requirements, poor circulating current, and capacitor voltage fluctuation suppression. First, this study proposes an adaptive MPC technique that adapts the number of candidate combinations to the steady and transient states, significantly reducing the computational burden. Second, an improved hybrid combination of an MPC with a proportional-resonance (PR) controller enhances the circulating current and capacitor voltage fluctuation suppression performance. According to the phase difference between the circulating current and the capacitor voltage, the circulating current and capacitor voltage can be suppressed at different times by changing the circulating current reference of the PR controller. The switching frequency can be reduced by using the PR controller’s output to adjust the input submodule number instead of changing the duty cycle. The proposed techniques were validated by simulations and experimental case studies with a three-phase grid-connected MMC.

Keywords

• modular multilevel converters
• model predictive control
• capacitor voltage fluctuation
• circulating current
• fast optimization