Advanced detection and localization of open circuit faults in two-level three-phase IGBT-based inverters using machine learning approaches and discrete wavelet transform

This article highlights the critical importance of safety and reliability of three-phase Insulated Gate Bipolar Transistor (IGBT) based inverters in industrial applications, especially regarding open circuit faults. These faults can lead to costly breakdowns and destabilize electric drive systems. Our approach is based on the use of advanced signal processing methods, including Discrete Wavelet Transform (DWT), combined with machine learning techniques such as Artificial Neural Networks (ANN), k-Nearest Neighbors (KNN), Support Vector Machines (SVM) and Random Forests (RF), for early detection and precise localization of faults. Our experiments demonstrated remarkable efficiency with a good classification rate of 100%, thus confirming their relevance in the context of two-level voltage inverters controlled by a Pulse Width Modulation (PWM) vector control strategy.