Space-vectors based hierarchical model predictive control for a modular multilevel converter

The modular multilevel converter (MMC) is a competitive candidate for medium/high-power applications, specifically for high-voltage direct current transmission systems. Model predictive control (MPC) is an advanced and flexible method for power converters. The existing MPC methods for the MMC 3-phase system treat whole system as a three independent single phase system, and the computational load increases geometrically according to the increase of the level of the MMC. This paper proposes a space-vectors based hierarchical model predictive control (HMPC) strategy for a 3-phase MMC system with independent cost functions. Three hierarchical mathematical models of the MMC are derived and discretized to predict the AC-side current, circulating current and capacitor voltage, respectively. By utilizing multilevel space-vectors and hierarchical model, the considered number of states can be reduced significantly with the highest DC voltage utilization ratio and good performance. In addition, this strategy doesn´t need the complex capacitor voltage sorting method and reduces the power loss by avoiding the unnecessary switching state transitions. The performance of the proposed strategy for 11-level MMC is verified through simulation results.