Characteristic Analysis Of Security State Assessment Data Of High-Speed Train Based On EEMD

In the process of train service for a long time, its key components on the bogie wear will lead to different levels of reduced performance, which seriously influence the train running safety and comfort, therefore it is particularly important to train safety warning and health maintenance. When high-speed trains run all around, the train sensors distributing in different positions will monitor vast amounts of real-time data. It has become the current hotpot and difficulty to extract features from the fault information of the real-time data and invert high-speed train health status. In view of the nonlinear and non-stationary of the monitoring data, this article introduces the ensemble empirical mode decomposition (EEMD) with the characteristics of adaption, and combines with other feature extraction methods to study the three working conditions:the rail wavy irregularity, the lateral dampers and anti-hunting damper performance degradation and different number of lateral dampers failure.(1) This section studied the related theory of the empirical mode decomposition (EMD).In order to solve the mode mixing phenomenon, the ensemble empirical mode decomposition(EEMD) was introduced, which assisted by noise. EEMD can not only decompose the signal in different time scales and obtain the local information of the signal, but also eliminate the mode mixing problem in a certain extent.(2) This section utilized the wavelet ridge method to extract the characteristics of the frequency and amplitude of the signal. EEMD combined with wavelet ridge method was proposed to detect rail wavy irregularity. Firstly, EEMD was carried out to pre-process the monitoring data, then used wavelet ridge method to extract the frequency and amplitude of the IMF, which contained more signal information. Judge whether it meets the characteristics of rail wavy irregularity, so as to achieve in time and frequency domain for detection of rail wavy irregularity.(3) Aiming at vibration feature of degeneration properties of lateral damper and anti-hunting damper, the empirical mode cross-correlation analysis method is proposed, and defining the empirical mode cross-correlation coefficient, adopting the method as a quantitative to describe the different reduced degree feature of damper. Using the method to research reduced degree simulating data of the two kinds of damper, and input experience of modal relations with each other to SVM to make identification. The experimental results proved the feasibility of this method.(4) In order to study local complexity characteristic of vibration signals of high speed trains, extracting and correlation dimension of permutation entropy and correlation dimension respectively from the IMF component which decomposed from EEMD, the complexity of the information on the available signals at different frequency bands is obtained. Taking ineffective working condition data of different number of lateral damper as analysis objection, the two kinds of method is used to extract features from the data. The results show that the two methods of feature extraction both can effectively describe the different fault characteristics of transverse vibration absorber, and the method based on EEMD permutation entropy is superior to which based on EEMD correlation dimension method.This work was supported by National Natural Science key Foundation of China. (No.61134002)...