DC Traction Current Network Fault Identification Algorithm

Many cities are facing a severe traffic congestion problem of surface public transportation in China, and the urban rail transit is an ideal solution with the advantages of large capacity, fast speed, three-dimensional and green Compliant etc. The driving power of Metro vehicle comes from electricity, then the traction network prone to failure and causes great consequence for normal operations. So it’s very important to adopt reliable DC traction feeder protection strategies to grantee the safe operations.In the paper, the several aspects of the research and analysis on the background of the DC traction network and the objector of the feeder current were done as follows. Firstly, the feeder current waveform characteristics of the various operating conditions were analyzed; Secondly, according to the comparison on the differences of mechanism and waveforms futures between the fault current of the traction network and the oscillating current, a novel feeder protection algorithm of the DC traction network based on correlation dimension was presented; Thirdly, the combination of wavelet packet and SVM algorithm was applied to identify the fault current and oscillating current, and found that the combination of algorithms is inadequate in extracting the low frequency oscillation characteristics of the DC traction network; finally, the conclusion that correlation dimension algorithm is superior in operation speed and reliability than the existing combination algorithm by comparing the two algorithms.According to the analysis results, the novel feeder protection algorithm of the DC traction network based on the correlation dimension not only has the advantages of high sensitivity and high reliability, but also is more suitable for the diagnosis of the DC traction network with complex operation states.