Model Predictive Current Control of a Three-Phase T-Type NPC Inverter to Reduce Common Mode Voltage

Abstract

This paper presents a reduced control set model predictive control (RCSMPC) method for three-phase T-type neutral-point-clamped (NPC) inverter. The whole control set (WCS) consists of all the 27 switching states of T-type NPC inverter. The reduced control set (RCS) with 19 switching states is formed from WCS by excluding the switching states with common mode voltage (CMV) value higher than one-sixth of input DC voltage [Formula: see text]. With RCS, single-objective model predictive current control method can restrict the CMV peak value to [Formula: see text]. To further reduce the CMV below this threshold, a cost function with the weighted sum of two control targets is formulated in the RCSMPC method. The two control targets of RCSMPC method are CMV mitigation and load current control. The weight for CMV is called bias factor. The RCSMPC method is computationally efficient, as the number of switching states is less than that of WCSMPC. To further reduce the computational burden, CMV values corresponding to all the switching states are calculated offline and stored in memory. Robustness of both the methods is investigated with parameter deviations at different bias factors and reference currents. The proposed method is validated using simulation and experimental results and compared with the existing methods.