Research On Fault Diagnosis For Gearbox On Morphological Component Analysis

The gearbox is a fault-prone part for its operating conditions and complex structure. Once when the gearbox failed, it will be the direct threat to running security of the urban rail train. Thus, the fault diagnosis for gearbox has a very important practical significance.The traditional hilbert envelope demodulation method is not suitable for analyzing issues of multi-component vibration signal’s modulation information, so the morphological component analysis(MCA) is applied for hilbert envelope demodulation. Using this method to analysis simulating vibration signal, the results show that this method can effectively separate the resonance component and impact component in signal, and highlight its characteristic information.The non-stationary and nonlinear signal usually has the large interfering noise and the frequency bandwidth in the actual measurement. If using MCA method directly, it will seriously affect the extraction of useful signal and fault feature information. Aiming at this problem, the combination of EEMD, MCA with energy operator demodulation is proposed in the fault diagnosis. Firstly, vibration signals of gearbox failure are decomposed by EEMD in this method; so the IMF component contains fault information can be selected according to each the IMF component’s correlation coefficient with the original signal; finally, IMF component selected do reconstruction; then the reconstructed signal is separated by the MCA; when each morphological component is obtained by MCA, then can get modulation fault information of the morphological component through the energy operator demodulation. Using this method to analysis the simulated signal and the actual fault signal, the results show that this method can effectively extract the characteristic information in vibration signal of the gearbox fault.In the view of order tracking prone to order aliasing in the analysis of non-stationary and variable speed signals, a method combined the order tracking technology based on a low-pass filtering with MCA was proposed. Using this proposed method to analysis a simulated variable speed signal and a vibration signal of a rolling bearing under changing rotating speed, the results show that the method of combining the improved order tracking technology with MCA can effectively extract the fault characteristic information of a vibration signal of a rolling bearing under changing rotating speed and avoid order aliasing. This proposed method also can reduce re-sampling the order and the amount of computation in the re-sampling procedure.