Fault Diagnosis for HVDC System Based on Wavelet Entropy Clustering and DS Evidence Fusion Theory

When an electrical accident happens, fault diagnosis of HVDC transmission system is the prerequisite step for isolating fault elements and restoring normal operation of the system, so an accurate and fast way of fault diagnosis is significant to improve the reliability of power system. Based on wavelet entropy clustering algorithm and DS evidence fusion theory, a new fault diagnosis method for HVDC system is proposed in this paper. This proposed method takes HVDC line voltage as monitoring variable and uses wavelet entropy to extract the features of the fault. Fault signals are clustered by fuzzy clustering algorithm considering the uncertainty and diversity of faults. Meanwhile, the norm weighted average is used to clear the allocation of basic reliability. The further precise classification of fault types is realized using DS evidence fusion and the decision-making method. The simulation model is built using both PSCAD/EMTDC and MATLAB software, and the correctness of the proposed method is verified by the two phase to ground fault on the AC side of the converter station.