Faults Detection Of Locomotive Traction System By Modified Wavelet Bispectrum Motor Current Signature Analysis

When faults emerge in locomotive traction systems,the emission signals will display more non-nonlinear and non-stationary properties than normal state.In order to find the phase relation in the fault signal and extract the fault feature,the modified wavelet bispectrum is introduced to motor current signature analysis.A novel locomotive traction system fault detection approach is proposed and the contributions are made as follows:The properties of the motor stator current under variable speed condition are analyzed.Mechanical faults induce the quadratic phase coupling increased in motor stator current signal.Since the phase coupling is changing with the motor speed,the modified wavelet bispectrum is used to process the motor current signal and extract the time-varied phase coupling.Using the wavelet bispectrum,some transient characteristic of the signal will be omitted.In the light of the problem,transient biphase random wavelet bispectrum is introduced.The advantages of transient biphase random wavelet bispectrum in nonstationary signal analysis and the time-varied quadratic phase coupling extraction is illustrated by computer-generated signal.Furthermore,the ability of fault feature extraction using the improved wavelet bispectrum is shown by test rig and locomotive online running test.Due to the fact that the faults induce the change of the distribution of the non-linear signal component,the wavelet bispectral entropy is presented to quantitatively describe the distribution of non-linear components in bifrequency domain.Since it is extremely sensitive to non-linear changes caused by mechanical faults,the wavelet bispectral entropy is employed to detect initial defects.The effectiveness and practicability of the proposed method is verified by tests.Based on the above research results,a fault diagnosis method for locomotive traction system is proposed.In this procedure,the wavelet bispectral entropy is extracted from the traction motor current signal to detect faults.Then,the transient biphase random wavelet bispectrum is employed to determine the location of the fault.The operation test of locomotives proved that the proposed method can detect the faults of traction systems effectively.