Use Of The Correlated EEMD And Order Tracking For Bearing Defect Detection Under Large Speed Variation

As important parts of train operation,bearings are prone to erosion and malfunction due to long-term heavy load and other factors.It affects the safety of train operation which makes the fault detection of bearings extremely necessary.Vibration diagnosis is widely used in the health monitoring of rolling bearings,currently available methods have been proposed mostly based on the assumptions of the constant speed or the speed with small variation.However,in practical applications,the REBs usually experience large speed variation especially in startup and shutdown,which cause the accurate fault diagnosis to be very difficult.Therefore,a fault diagnosis method based on correlation EEMD and order tracking is studied in this paper.The details and research results are as follows.Firstly,the basic structure of the bearing and the vibration mechanism of the bearing fault is studied,the characteristics of vibration signals under different faults are also introduced.In addition to a single fault,it also involves a compound fault,the simulation model is obtained,and applied to the simulation verification.Secondly,a method of extracting instantaneous rotation frequency(IRF)of bearing with CEEMD is proposed.In the method,the vibration signal is first decomposed into a series of intrinsic mode functions(IMFs)with the EEMD,and correlation analysis is subsequently conducted between the selected IMFs and original signal,through correlation analysis,the shaft IMFs could be extracted.To highlight IRF,take the reciprocal of correlation coefficients of shaft IMFs as weight in reconstructed process,the IRF can be estimated accurately from time-frequency representation of reconstructed signals with peak search and curve fitting.Finally,a fault recognition algorithm based on time spectrum kurtosis and peak coefficient is proposed.In this paper,the time-spectral kurtosis algorithm is used to locate the structural resonant frequency excited by the fault signal,according to the resonance frequency range to determine the selection of fault IMF,then the envelope order spectrum of the fault IMF is obtained.A peak coefficient is established by setting a reasonable threshold to find the richest diagnostic information for final decision making.The simulation and experiment verify the effectiveness of the algorithm.The research results are of great significance to engineering application of vibration fault diagnosis technology for bearing faults.