基于STDR/SSTDR阻抗特征的IGBT故障诊断研究

Reliable fault detection of insulated gate bipolar transistors (IGBTs) is essential to enhance system reliability and performance in power electronic applications. In this paper, an advanced sequence/spread spectrum time domain reflectometry (STDR/SSTDR) diagnostic method is proposed for accurate extraction of impedance signatures that reflect the time-domain response of the device under different operating states. This method integrates cross-correlation analysis with multivariate linear regression to extract principal components from complex reflected waveforms, thereby enabling simultaneous identification of spatial location and impedance signatures. Moreover, to enhance the discrimination of device fault states, a dual-regulation scheme involving signal frequency and DC bias is introduced. High-frequency SSTDR signals are utilized to enhance the identifiability of primary reflection peaks, thereby enabling precise determination of the relative position between the target device and the test terminal. Low-frequency positive-bias excitation helps suppress parasitic effects, enabling reliable retrieval of the impedance signatures of IGBT chips in their conducting state. Experimental results demonstrate that the proposed method achieves high accuracy and robustness in detecting open-circuit (OC) faults across various types of IGBTs, with particularly strong stability and adaptability under conditions of severe signal overlap and complex device structures.