Study On Bearing Fault Detection Method Of EMU Mechanical Transmission System

With the continuous emergence and speeded up the situation of Railway High-speed EMU, accurate detection of train motor bearings running state is very important. At present, the rolling bearing fault diagnosis technology mainly include stemperature analysis, vibration analysis, the oil sample analysis, noise analysis method, acoustic emission and vibration diagnosis, as a common natural phenomenon, vibration contains a wealth of abnormal or fault information in the project, Vibration characteristic is the important means to judge the operation status of the equipment, therefore vibration signal monitoring is commonly used to determine the equipment operating status.Firstly, this paper introduces the EMU fault forms, causes and the evolution of bearing fault diagnosis, these methods include time domain parameter method, the traditional Fourier analysis, wavelet transform based on Hilbert transform, wavelet packet transform based on Hilbert transform, wavelet packet analysis method based on correlation coefficient, modal decomposition method based on correlation coefficient, ensemble empirical mode decomposition method based on correlation coefficient.Then, using Matlab simulation verify the accuracy of each method, ultimately selected ensemble empirical mode decomposition method based on correlation coefficient as a last judgment by experiments. This method not only overcome the traditional Fourier analysis in time and frequency resolution does not take into account the resolution of the problem, Wavelet analysis method relies heavily on the band wavelet function and the problem of uneven distribution,but also to overcome the empirical mode decomposition modal aliasing problems. Is far more satisfactory method of diagnosis.Finally, the program is transplanted to LabVIEW, written in LabVIEW real-time monitoring program and accomplish the real-time diagnosis. This system contains real-time signal acquisition unit, time domain diagnostic unit, frequency domain diagnosis unit, integrated fault determination unit and history data recording unit. The part of signal processing based on mixed programming of LabVIEW and Matlab, processing and analysis of the bearing failure vibration signal by Matlab Script node called Matlab procedures, and display the results by user interface, realizes the data acquisition, analysis, processing, display and diagnosis. Algorithms studied in this paper can be applied to the system state detection and fault diagnosis system to achieve predict potential failures early in the strong weak background noise and provide a theoretical basis for the study of fault diagnosis system, fault diagnosis is important in the field of machinery and equipment value.