Data-driven fault diagnosis of power converter systems

Abstract

As power converters advance, their significance in daily life grows ever more pronounced. Ensuring the reliability and safety of power converter operation necessitates prioritizing fault diagnosis for power converters. In this paper, a data-driven fault diagnosis method for power converters is proposed, which is aimed at the exhaustion of IGBT open circuit faults in power converters. Meanwhile, the imbalance in the actual project is also taken into account. By comparing the accuracy and diagnosis time of four different data-driven diagnostic methods, it is found that Random Vector Functional Link (RVFL) has the shortest diagnosis time and high accuracy, indicating that this method has the best effect on IGBT fault diagnosis in the face of sample imbalance.