Research On On-line Monitoring For Partial Discharge Of High-speed Railway High-voltage Cables

Normal speed railway supply line directly to the catenary from the traction substations,transmission using the laying of overhead lines.The technical and safety of traction power supply high-speed railway rail relatively high normal speed,which is usually underground cable laying way,so that the cable is often difficult to identify an unsafe condition,and a partial discharge is an important manifestation of cable insulation deterioration condition,cable insulation performance degradation can result in significant traction network security risks arise,in order to detect serious cable insulation degradation,ensure the safe operation of the traction power supply,we select local discharge method for traction power supply high voltage cable online monitoring a study.First,the reasons discussed partial discharge,partial discharge mechanism discussed cables,capacitors using three models,discharge process was analyzed from the principle;described partial discharge of several common types of defects,analysis of the discharge characteristics under different defect;ATP application software to build high-voltage cable traction simulation model of partial discharge,partial discharge and the fault module is introduced inside the cable,as proved the deepening insulation deterioration,partial discharge magnitude cables also increase results.Kind with a cable,in a high pressure lab building physics simulation model validated ATP software simulation results,and to provide a sample cable partial discharge pattern recognition.Secondly,for the type and characteristics of the cable operating conditions of noise,select the EMD algorithm and wavelet analysis to white noise and periodic narrowband interference noise elimination were studied.Combine the advantages of the two algorithms is proposed based on the acquisition of EMD wavelet threshold filtering method,simulation and experimental data PD to the filtered signal experiments show that the method has a better filtering effect.The online monitoring of the cable insulation status includes the identification of the type of partial discharge.The input object of pattern recognition is the characteristic quantity extracted from the cable partial discharge signal.According to the characteristics of the cable partial discharge signal,this paper extracts the four characteristic parameters of steepness,cross-correlation factor and discharge phase,and applies improvements respectively.BP neural network algorithm and RBF neural network algorithm to identify three kinds of defects.The results show that the RBF recognition algorithm can effectively solve the shortcomings of BP network local optimization and slow convergence speed,and the overall recognition rate is relatively high.The online monitoring of the cable insulation status includes the identification of the type of partial discharge.The input object of pattern recognition is the characteristic quantity extracted from the cable partial discharge signal.According to the characteristics of the cable partial discharge signal,this paper extracts the four characteristic parameters of steepness,cross-correlation factor and discharge phase,and applies improvements respectively.BP neural network algorithm and RBF neural network algorithm to identify three kinds of defects.The results show that the RBF recognition algorithm can effectively solve the shortcomings of BP network local optimization and slow convergence speed,and the overall recognition rate is relatively high.