Open-switch fault diagnosis system based on one current sensing for a matrix converter using neural network

It is difficult to diagnose a three-phase matrix converter using a mathematical model, because a matrix converter consists of nine switches with various nonlinear factors. Since a neural network does not require any mathematical system model, it is a suitable technique for fault classification in matrix converters. In this manuscript, a fault diagnostic system for three-phase to three-phase matrix converters using a neural network is proposed to detect a fault and identify its location. The proposed diagnostic system can detect faults using just one phase current waveform which is very efficient in terms of cost of sensors and system complexity. This method was evaluated using simulation and experimental data sets in two scenarios. The results confirm that in different normal and abnormal situations the system achieves performance levels in excess of 98%.