An improved EWT method for fault diagnosis of rolling bearings
Considering the problem of empirical wavelet transform(EWT)in extracting optimal frequency band of the rolling bearing fault signal,an improved EWT method based on extracting energy envelope trend line to adaptively divide frequency band was proposed and applied to rolling bearing fault diagnosis.The Teager energy operator was used to convert the spectrum into energy spectrum,and the energy envelope was obtained by repeated Hilbert transform.Local maximum values were extracted and smoothed to obtain the energy envelope trend line,and the first-order difference was performed to select effective extreme points to adaptively divide the frequency band.A normalized fault characteristic frequency saliency index was constructed as an effective criterion for fault diagnosis and optimal resonance frequency band selection.The algorithm was verified by rolling bearing fault simulation and experiment data.The results showed that compared with the original EWT,the proposed method can effectively identify the early faults of rolling bearings and reasonably select the optimal resonance frequency band.The proposed indexes for the outer and inner race fault data can be increased by 48.0％and 174.1％on average.
rolling bearingsempirical wavelet transformfault diagnosisresonance demodulationadaptive signal decomposition
NETL
NSTL
万方数据
