Fault Identification of Winding Axial Displacement and Inter-turn Short Circuit for UHVDC Transformer

Converter transformers are key primary equipment for UHVDC system. The nonlinear switching actions of thyristors seriously deteriorate the working conditions of station converters and make the probability of outage due to component aging remain high. In view of the deficiency of traditional fault diagnosis of converter transformers which mostly relies on non-electric quantity monitoring, this paper presents a method for fault identification of winding axial displacement and inter-turn short circuit based on transformer electrical waveforms by establishing a converter transformer lumped model and analyzing field recorded wave data. Firstly, the accuracy of the converter transformer’s simulation model is verified by comparing the simulation waveforms with the field recorded waveform data, which lays a foundation for obtaining usable fault data sets. Then, the fault features in converter transformer's electrical waveforms were extracted by elliptic equations, and the support vector machine (SVM) was constructed to judge the fault existence and identify the fault types. The test results show that the accuracy of the proposed method is higher than 70%. The effectiveness of transformer fault diagnosis using UHVDC field recording data is verified.