Multi-resolution analysis for converter switch faults identification

Multilevel inverters (MLI) are extensively used in various fields and industrial applications. This study proposes a simple algorithm for detection and identification of switch faults in insulated gate bipolar transistors (IGBTs) of multilevel inverter feeding an induction motor drive. The effects on the system for switch shorted are more severe than the switch opened if no protective action is taken in time. A fast and novel fault identifier designed in this study is used not only for switch short fault but also for switch open fault by analysing the performance of MLI drive system using wavelet based multi-resolution analysis (MRA). On occurrence of switch fault in inverter, the fault information appears in the signals in frequency region resulting in variation in the details of level of wavelet transform (WT). The instant of variation in the WT is correlated with the instant of fault. The voltage and current signals from the experimental setup were used as input for wavelet analysis to develop suitable feature vector that will act as signature to discriminate faulty system from healthy one. The experimental prototype of five level diode clamped multilevel inverter feeding an induction motor has been implemented to validate the fault analysis with simulated result.