Abstract: Localized faults, such as spalling and crack, in rotating machinery parts tend to result in shocks and thus arouse transient impulse responses in the vibration signal and thus present a potential approach for fault feature extraction. Based on transient modeling, a method combining with least square method is proposed and applied to iteratively identify transient parameters. Based on Morlet wavelet parametric expression, a double-side asymmetric transient model is firstly built; then, Levenbery-Marquardt method is introduced to identify parameters of the model. With the transients extracted from the signal, Wigner-Ville distribution is applied to show high resolution and no cross item time-frequency representation of transients. The transient parameter identification method based on LSM is used to extract feature of a faulted bearing, and the results show that the transients is obtained through the proposed method and eventually time-frequency feature of the fault is well expressed in a high resolution and no cross item form.

Key words: Bearing fault diagnosis, Levenbery-Marquardt method, Parameter identification, Transient signal