Fault Diagnosis of Autotransformer Winding Axial Displacement Using FRA Method Based on Dynamic Frequency Division

Aiming at the problem that the traditional frequency response analysis (FRA) method could not identify the common axial displacement fault of autotransformer winding, the FRA method based on dynamic frequency division was proposed. Firstly, a simulation platform of autotransformer axial displacement fault was built to obtain the winding frequency response data under different axial displacement faults. Secondly, the peak point of the amplitude-frequency characteristic curve and the zero-crossing point of the C-L phase-frequency characteristic curve with the same frequency were used as the preliminary frequency division points. Meanwhile, the frequency response data was dynamically divided into multiple frequency bands and the polar plot with frequency division was drawn. Then, the texture features of the 4 gray-gradient co-occurrence matrix and the corresponding normalized characteristic parameters of the polar plot were extracted. Finally, the state of the autotransformer winding was analyzed by the changing law of the graphs and features. Experimental results show that using the FRA method based on dynamic frequency division, the overlap data points generated from polar plot is improved, which is conducive to graphic analysis and feature extraction. When axial displacement faults occur in different windings, the polar plots in each frequency band have obvious changing trend. When the same winding has different degrees of axial faults, the difference of polar plots tends to become larger as the increase of fault degree. The FRA method based on dynamic frequency division can accurately distinguish the axial displacement fault winding and the fault degree of autotransformer, and can provide a reference for the on-site fault diagnosis of autotransformer. © 2022, Editorial Department of China Railway Science. All right reserved.