Solving harmonics reduction problem in multilevel inverters by switching neural networks-based approach

Abstract

Multilevel inverters are increasingly considered in integrated generation systems (IGSs) due their significant advantages and low voltage output harmonics content. In this paper a switching neural network (SNN)-based method to provide the switching pattern for harmonics reduction is proposed. We perform the implementation of this new paradigm to establish the switching pattern for a five level three-phase inverter. The provided results as well as the main guidelines for SNN software and hardware implementation show the significance of this approach.