The collected vibration signals from defective rolling element bearings are generally non-stationary and corrupted by strong background noise. The weak fault feature extraction is crucial to mechanical fault diagnosis and machine condition monitoring. A new method EWT-based (Empirical Wavelet Transform) for bearing fault diagnosis is proposed in this paper. It consists of four parts. Firstly, the frequency ranges of meaningful modes are self-adaptively obtained by combining scale-space representation and Otsu’s method. Secondly, the meaningful modes are acquired by utilizing EWT to decompose the raw vibration signal. Thirdly, the first two modes possessing maximum kurtosis are selected as fault components. Lastly, the fault-related features could be observed in the time domain and envelope spectra of the selected modes. Experimental results verify that the proposed method is very effective for bearing weak fault diagnosis and the performance of proposed method is obviously better than the method of empirical mode decomposition (EMD).

This content is only available via PDF.
You do not currently have access to this content.