Open-Circuit Faults Detection in Three-Phase Inverters Using PCA and WE-SVM Algorithms

Open-circuit (OC) faults are a frequent issue in the switching devices of power electronic converters, particularly in three-phase inverters. These faults can lead to significant performance degradation and, if undetected, may damage other system components. Accurate and timely detection and precise localization of the faulty switch are essential to ensure system reliability and enable effective fault-tolerant operation. This paper presents a combined approach for detecting and localizing OC faults in three-phase inverter switches. The Principal Component Analysis (PCA)-based method is employed for early fault detection. At the same time, the faulty switches are localized using a Support Vector Machine (SVM) classifier, utilizing Wavelet Entropy (WE) features derived from three-phase current signals. The proposed PCA and WE-SVM-based method demonstrates improved accuracy and reduced detection time compared to existing techniques. The effectiveness of the work is validated using MATLAB Simulink software, which shows strong alignment with theoretical predictions.