Optimised Model Predictive Control for MMCs: Enhancing Performance and Reducing Computational Burden

This paper proposes an Improved Folding Model Predictive Control (IFMPC) strategy for Modular Multilevel Converters (MMCs) that significantly reduces computational complexity while enhancing control accuracy and real-time feasibility. Although a Finite Control Set MPC (FCS-MPC) uses actual capacitor voltages, the state-of-the-art methods rely on averaged capacitor voltages for prediction, which can introduce inaccuracies. The proposed IFMPC consistently employs instantaneous capacitor voltages to generate voltage vectors during the prediction stage, thereby reducing prediction errors and enhancing both transient and steady-state performance. The proposed approach integrates four critical control objectives AC current tracking, circulating current suppression, arm energy balancing, and leg energy distribution into a unified cost function with only two tunable weighting factors, simplifying the tuning process without compromising robustness. Real-time hardware-in-the-loop (HIL) validation on a scaled MMC prototype demonstrates rapid dynamic response, effective disturbance rejection, and reduced total harmonic distortion (THD) under various operating conditions, including parameter mismatches and grid harmonic distortions. Comparative analysis against existing indirect MPC techniques reveals that IFMPC achieves superior control performance with reduced computational burden, making it well suited for MMCs with a high number of submodules (SMs). The proposed method offers a scalable and industry-ready solution for advanced MMC control in high-power applications.