Spurious resistance identification in SPWM inverter circuit feeding power to induction motor

Abstract

This paper proposes a simple yet effective method to identify faulty switch and faulty phase in Sinusoidal Pulse Width Modulation (SPWM) inverter feeding power to induction motor. A typical but frequently appearing fault in inverter switch may occur due to the introduction of spurious resistance in series with any of the IGBTs. Such fault normally gets introduced due to poor contact point, aging, overheating and in most of the situations, these kind of faults remains undetected at their early stage by the drive protection system. If such fault remains undetected, due to overstress & overheating in healthy IGBTs, may lead to failures in other components as well. Generated heat in switch contact point accelerates the deterioration of the switch performance as the generated heat would lead to further increase of contact resistance. Therefore a positive feedback type situation arises and that may lead to a catastrophic failure. This paper proposes an algorithm developed in MATLAB to detect such faults and faulty phase in SPWM inverter by carrying out analyses on motor currents captured from a model equivalent to such situations implemented in PSIM simulation software. Some features have been extracted using Fast Fourier Transforms (FFTs) on Park's Vector Modulus (PVM) of three phase motor current profiles & finally Simplified Fuzzy Art-Map (SFAM) classifier has been employed to detect faulty switch and degree or level of fault. The obtained results provide a satisfactory level of accuracy.