An Improved Reduced Complexity Model Predictive Current Controller for Grid-Connected Four-Leg Multilevel Inverter

Abstract

Model predictive current controller offers numerous benefits over the traditional controllers in grid-tied inverters. However, the real-time execution duration reduction of model predictive controller in multilevel inverters is a big challenge, which has been investigated by several researchers. The work presented in this article exploits the three-dimensional space vector representation of a four-leg neutral point clamped three-level inverter in a-b-c coordinates to lower the real-time execution duration of model predictive controller. The objective function determination counts of model predictive controller in the process of controlling neutral point voltage and source current of four-leg neutral point clamped inverter are significantly reduced in the proposed scheme. Therefore, real-time execution duration is reduced as well. In the proposed scheme, three virtual mutually orthogonal sliding planes decide the objective functions to be determined. This method evades the use of any weighing factor in the objective function, which completely obviates the parameter tuning requirement for the control of the four-leg inverter. The extension procedure for higher level four-leg inverter is also explained. Experimental evidence is presented to prove the superiority of the core idea with respect to a recent work.