Power Switch Fault Diagnosis Technique for Three Phase Voltage Source Inverters

Abstract

This study proposes an automated fault diagnosis method for switch faults in three-phase inverters, based on current waveform feature extraction. The research focuses on analyzing three-phase output current waveforms under different fault conditions, including open-circuit fault (OCF), short-circuit fault (SCF), gate driver fault (GDF), and aging (AGN). Multiple key indicators such as average current, root mean square (RMS) current, peak-to-peak value, and Park’s current are extracted, and two normalized current indices are proposed as the basis for fault identification. The proposed approach can be seamlessly integrated into existing current sensors and control algorithms, requiring only minimal software adjustments. Moreover, the method is robust and applicable under varying conditions consistently achieving high diagnostic accuracy.