Defect Diagnosis for Autotransformer Winding in High-Speed Railway Based on Transfer Function Pole Distribution

Winding defect in autotransformer (AT) significantly affects the safety of traction power supply system. Frequency response analysis (FRA) is a widely used method to diagnose the winding status. Since the special structure of AT, it is challenging to detect specific faulty windings accurately. This article presents a defect diagnosis method for AT based on the transfer function pole distribution. First, a test platform for AT is established to obtain the FRA data on axial displacements (ADs), series capacitance variations (SCVs), and an interdisk short circuit (SC), as well as mixed defects. Next, the pole distribution diagram is expressed in the form of polar coordinates, considering the amplitude–frequency and phase–frequency curves and incorporating the fast relaxed vector fitting method. Then, the feature parameter combination for analyzing defect windings and defect types is defined using the gray correlation analysis results in conjunction with the Tamura textural features and gray-gradient co-occurrence matrix (GGCM). Finally, the feature parameters are used as input data to the cuckoo search-support vector machine (CS-SVM) for the classification of faulty windings and defect types. The results show that the proposed method can accurately identify the windings and types of both single and mixed defects.