Traction Network Fault Identification Of The Through-type Co-phase Traction Direct Power Supply System

High-speed railway has the advantages of high speed,large conveying capacity,low energy consumption per unit,good safety,good economic benefits,and comfortable and convenient.Therefore,it has become the key development direction of China's science and technology and economy.In recent years,most of China's railways have been electrified,so the traction power supply system in electrified railways has become the basic condition for safe,stable and high-speed operation of high-speed railways.The traction power supply system of high-speed railway has complex structure and poor operating conditions.The cold and warm weather and lightning weather will bring great obstacles to the normal operation of the high-speed railway traction power supply system.Short-term high-amplitude lightning will cause the traction substation to trip or even The broken line of the traction network makes the train power failure unable to operate normally,which seriously affects the safety and stability of the high-speed railway.Therefore,the research on the fault of the traction network has become an important link to ensure the normal operation of the railway.Among them,fault identification plays an important role in fault research.Its significance lies in how to guide the protection,that is,it can distinguish the faults and can make corresponding protection measures to avoid the phenomenon of refusal and misoperation.This paper analyzes and identifies three kinds of disturbances such as lightning strike fault,lightning strike interference and ground fault in the traction network of high-speed railway continuous cophase traction direct power supply system:(1)Based on the simulation model of the high-speed railway through-phase inphase traction direct power supply system,according to the line structure of the traction network,the lightning strike fault,lightning strike interference and ground fault that may occur in the traction network of the direct power supply system are modeled and passed.Simulation experiments show three transient characteristics of the disturbance.(2)For the distinguishing problem of lightning strike fault,lightning strike interference and ground fault of the traction network of the through-phase in-phase traction power supply system,the time-domain and frequency-domain analysis of the traveling wave of the fault transient current,using the integral feature method in time domain analysis,the wavelet energy ratio analysis method in the sample entropy method and the frequency domain analysis distinguishes three kinds of disturbances.It can be seen from the experimental results that although the above three methods are simple and intuitive,the single criterion set has certain limitations in the fault differentiation.(3)A single criterion set for the fault transient current traveling wave time domain and frequency domain analysis method can not distinguish the lightning strike fault,lightning strike interference and ground fault of the traction network,and proposes the Modified Ensemble Empirical Mode Decomposition(MEEMD)and Probabilistic Neural Network(PNN)intelligent identification method.The MEEMD decomposes the fault signal to obtain the Intrinsic Mode Function(IMF).The sample entropy and permutation entropy are used to extract the feature of each IMFs component.Combined with PNN,the fault recognition experiment is carried out.The experimental results show that MEEMD and permutation entropy are based.The PNN intelligent identification method can better identify the lightning strike,lightning strike and ground fault of the traction network,but the recognition rate still needs to be improved.(4)There is still room for improvement in the fault recognition rate obtained when the PNN identifies the lightning strike fault,lightning strike interference and ground fault of the traction network,three kinds of disturbances are identified by the Support Vector Machine(SVM)which has obvious advantages in processing small sample data and nonlinear sample data.MEEMD decomposes the fault signal to obtain the IMFs component,extracts the feature of each IMFs with sample entropy and permutation entropy,and combines SVM to carry out fault recognition experiment.It can be seen from the experimental results that the SVM intelligent recognition method based on MEEMD and permutation entropy can be accurate.Identifying the lightning strike,lightning strike and ground fault of the traction network,the fault recognition rate is significantly higher than the fault recognition rate of the PNN.