Open-Circuit Fault Diagnosis of IGBT Devices in Traction Inverters Based on BL-RF

The traction inverter converts direct current (DC) into three-phase alternating current (AC) to drive the traction motors. The Insulated Gate Bipolar Transistor (IGBT), as the core power device in the traction inverter, plays a crucial role in determining the efficiency and stability of inverter operation. To address IGBT open-circuit fault diagnosis, this paper proposes a novel method based on the integration of Benford’s Law (BL) and the Random Forest (RF) algorithm, referred to as BL-RF. Specifically, the three-phase current signals of the traction motor are used as the primary source for fault characterization. After preprocessing, the occurrence probabilities of the leading non-zero digits in each phase current are calculated to construct the diagnostic feature vector. Subsequently, the Random Forest classifier is applied to identify and locate IGBT open-circuit faults based on the extracted features. Furthermore, an experimental dataset is constructed using a traction system test platform with simulated IGBT open-circuit faults. Experimental results demonstrate that the proposed BL-RF method achieves an accuracy of up to 95.35% in identifying IGBT open-circuit faults, which confirms the effectiveness and feasibility of the approach.