Mean Normalized Multi-objective Predictive Control for Matrix Converters with Common-mode Voltage Elimination

Abstract

This paper presents a method to realize multiple objectives of sinusoidal source and load currents in a direct matrix converter (DMC) fed system using finite control set model predictive control (FCS-MPC) without any weighting factors in the cost function. Moreover, this is achieved while maintaining zero common-mode voltage (CMV) by utilizing only those switching states of the DMC that switch three different input phases at the load terminals in each switching state. Further, the input power factor is also maintained near unity. The absence of any proportionality factors corresponding to the errors in source and load currents in the cost function is the result of a scaling algorithm based on statistical mean normalization that automatically adjusts the errors to a directly comparable scale before implementing a switching state. Simulation and experimental results are presented in this paper validating the proposed method of control and comparing its performance with an optimum weighting factor based predictive current control scheme for a three-phase to three-phase DMC.