Comparative Analysis of Predictive Control Systems Applied to a Grid-Tied NPC Inverter

Abstract

In this paper, two control methods based on predictive technique have been proposed for NPC grid tied inverter. The first method is the well-known Finite Set Model Predictive Controller (FS-MPC). It deals with the finite number of the converter switching states that optimizes a predefined cost function. The absence of the modulator in steady state performance results in generating variable switching frequency increasing system losses. Model Predictive Control with constant switching frequency (MPC-CSF), is introduced and discussed. The algorithm of MPC-CSF works in finding the optimal angle of the voltage vector in the first loop and then in finding the optimal amplitude of the voltage vector in the second loop. Simulations for the two proposed concepts were performed on Matlab/Simulink in order to verify their performance. Both methods are compared with respect to their complexity, simplicity, ability to inject current with low THD, dynamic response, and capacitor voltage balance.