Fault diagnosis of cascaded H-bridge multilevel inverter based on GWO-VMD

In order to accurately and efficiently identify single and double tube faults in cascaded H-bridge multilevel inverters, a model-based fault diagnosis and identification method is proposed. The collected three-phase current data were first subjected to rigorous data pre-processing. The pre-processed data were then subjected to noise reduction, feature decomposition and reconstruction using the Variational Modal Decomposition (VMD) algorithm enhanced by the Gray Wolf Optimization (GWO) strategy to extract key signal features. Fault characterisation of the reconstructed signals by the computational teager operator (TEO) is also investigated, Accordingly, the waveform state was determined, and the characteristics of the relevant points were captured. The value was then processed numerically to a certain extent, and the failure mode was recognized. Simulation results show that the accuracy of inverter fault identification using this method is high, reaching about 98%.