A Hybrid Model Predictive Multilayer Control Strategy for Modular Multilevel Converters

This paper presents a novel hybrid model predictive multilayer control (HMPMC) strategy for modular multilevel converters (M2LCs). The proposed hybrid strategy uses a modified model predictive direct slope control (MPDSC) concept to regulate the load current while concurrently minimizing the switching frequency, circulating current, and capacitor voltage variations through a new MPMC concept. The multilayer philosophy uses dedicated control layers to independently evaluate its control variables. As the main advantage, the hybrid strategy enables operation of M2LC at low switching frequencies. Moreover, this paper proposes a hierarchical screening algorithm, which estimates the minimum number of switching states required to be evaluated for each control variable, reducing the computational burden. An execution order for each control layer, in which the variables are assigned to control layers as appropriate, is also incorporated to further reduce the computational burden. Instead of using conventional weighting factors, this paper uses tuning factors for each layer, which is based on the number of switching states of the lowest cost to make the tuning process efficient and adaptable to any operating conditions. To demonstrate the validity, a prototype singlephase, five-level, 380-VA M2LC is designed, built, and controlled through the proposed strategies. The performance of prototype is investigated under both the steady-state and transient conditions, and the experimental results are presented in comparison with simulations and benchmarked against both the conventional finite control set model predictive control (FCS-MPC) and an improved MPC strategy. The results clearly indicate that the proposed strategy enables efficient operation of the converter at both significantly lower switching frequency and reduced computational burden, which is preferred for high-power applications.