Fault Diagnosis Method for Bearings Based on Morphological Construction of Scale Space and Spectral Kurtosis
In response to the problems of harsh service environment and difficult fault feature extraction of high-speed train axle box bearings,an empirical wavelet transform(EWT)fault diagnosis method is proposed based on morphological construction of scale space representation.Firstly,a scale space representation method is proposed based on morphology to achieve the adaptive recognition of resonance frequency band boundaries.Secondly,combining scale space representation based on morphology with spectral kurtosis to identify the resonance frequency bands.Then,a filter bank is constructed based on identified frequency band boundaries to adaptively decompose the signal.Finally,the envelope spectrum analysis is performed on decomposed signal to identify the fault feature frequencies and achieve the bearing fault diagnosis.The vibration test results of high-speed train axle box bearings show that the proposed method can effectively improve the accuracy of frequency band division and computational efficiency,and accurately diagnose the bearing faults.
rolling bearingfault diagnosiswavelet transformmorphologyscale spacekurtosis
NETL
NSTL
万方数据
