Wavelet decomposition based fault detection in cascaded H-bridge multilevel inverter using artificial neural network

Multilevel Inverter has been becoming extremely popular in modern power system. Rapid expansion of renewable energy sectors, transformer less grid synchronization and demand for harmonic free power quality are such igniting factors for its popularity. As power electronic devices are prone to various faults, hence to ensure the stability and reliability of system operation, this paper presents an intelligent fault detection method for open switch failure in a five level cascaded inverter. Wavelet decomposition is used to extract valuable features from the output voltage signals of the inverter. Artificial Neural Network is employed at last for the switch fault detection purpose. Short circuit faults are converted to open circuit faults by incorporating a fuse before each switch. Simulation results confirm the efficacy of the fault diagnostic approach with utmost of 99.9% classification accuracy.