Fault detection of switches in multilevel inverter using wavelet and neural network

Abstract

A Multilevel Inverter (MLI) with an advantage of low Total Harmonic Distortion (THD) is appropriate for the grid integration of solar photovoltaic power plants but the high number of switches makes them prone to failures affecting their reliability. In this paper a comprehensive outlook of fault diagnosis method along with practical implementation of 5 level Cascaded H-Bridge Multilevel (CHBMLI) is done. An algorithm is developed based on the output of CHBMLI. The operating range of input voltage to CHBMLI for proper working of algorithm is determined and an attempt is made to determine the operating range for a neural network based fault detection.