Wavelet based component fault detection in diode clamped multilevel inverter using probabilistic neural network

Multilevel inverter topology are becoming extremely important now-a-days due to the rapid growth in the renewable energy sector. Any kind of fault in the inverter directly hampers the normal operation of smart grid. In this context, this paper presents an intelligent technique for switch fault detection in diode clamped inverters for mid power application. It is necessary to detect and locate fault location and remove it quickly as reliability of power electronics components are important for stable operation of power system and electric drives. In this paper probabilistic neural network with discrete wavelet transform as signal preprocessor is employed for the fault detection in a three phase three level diode clamped inverter fed to LC load. Simulation results indicate highly satisfactory result with utmost of 99.8% fault detection accuracy.