Research On Fault Diagnosis Of High-Speed Train Based On Information Fusion Of Process

In the recent years in China, the development of high-speed rail has entered the peak period. The security state during the service of high-speed train becomes the core content of performance monitoring system. The security state is changed in the process of long-term service. The signals of sensors, which are installed at key locations, collect rich fault information. Then, taking appropriate information processing methods can effectively realize the inversion and identification of high-speed train running state. In connection with the signal of the most commonly occurred situations:the normal situation, the eight kind locations of anti-hunting demolition faults, the four kind locations of lateral damper demolition faults, and the two kind locations of air spring demolition faults, added 15 kinds of working conditions, a fault diagnosis system named information exetgy, which can effectively realize fault identification and fault orientation, was established. This system was based on process information fusion.The following study is completed:1) Compose simulation model and generate simulation data. With the software of dynamic simulation SIMPACK, for train model, set different working parameters, generate various types and locations of simulation data.2) Fault identification. The paper studys the implication and applicability of 4 multi-domain information entropys in the fault identification of high-speed train bogie. Combining information fusion theory, the paper gives four kinds of dual-channel information entropys, which effectively realize the fault identification of 4 kinds of fault types, including normal situation, the eight kind locations of anti-hunting demolition faults, the four kind locations of lateral damper demolition faults, and the two kind locations of air spring demolition faults.3) Fault orientation. The shortcoming of the state-based fault diagnosis methods is pointed out, and a fault orientation method of information exergy based on information of process is proposed. The paper gives a fault diagnosis model of information exergy feature extraction-feature selection-dimensionality reduction-pattern recognition. Firstly, connected with the faults of same typle but different location, the multi-sensor information exergy sample matrix based on the time domain singular spectrum entropy are calculated respectively, and high-dimensional feature vectors are constituted. The simulation results showed the effectiveness of multi-point information exergy features. In order to eliminate redundant features and bad features, a method based on ReliefF, Fisher ratio, Mahalanobis distance and ISOMAP algorithm is established. The results showed that this method can effectively reduce the dimensions of features, and the identification rate of fault orientation is improved.4) A high-speed train bogie fault diagnosis system of information exergy was established,which was based on proeess information fusion. This system can effectively realize fault identification and fault orientation. The diagnosis approaeh is:Firstly, identificate the sort of vibration signal by the method of multi-domains entropy feature. Then, determine the orientation with the method of multi-point information exergy features. The results show that the diagnosis system was a powerful model of fault diagnosis in the area of high-speed train diagnosis.This work was supported by National Natural Science Foundation of China--Key promblem research on high-speed train safety assessment based on data(No.61134002).