Study on Keypoint Extraction Method of Phase-Resolved Partial Discharge Pattern in Power Transformer

During the manufacture and installation of power transformers, or after long-term operation, there may be various partial discharge (PD) sources inside the power transformer. In order to accurately assess the insulation state, it requires accurate measurement and diagnostics of type and location of PD source. Therefore, it is necessary to find an accurate method to extract effective features from phase-resolved partial discharge (PRPD) patterns, which can effectively characterize the PD type. In this paper, four keypoint extraction algorithms (the scale invariant feature transform (SIFT), the speeded up robust features (SURF), oriented fast and rotated brief (ORB) and BRISK methods) are introduced. For different types of PDs, different methods were used to extract a fixed number of key points. Then, the key points of the PRPD pattern of same defect type were matched. Thus, keypoint extraction methods were compared according to the degree of matching. According to the matching results, the ORB algorithm and the SURF algorithm have better performance, and the SIFT has the worst result.