Oscillatory behavior based fault feature extraction for bearing fault diagnosis

An intelligent fault signature extraction scheme based on oscillatory behaviors is reported in this paper for bearing fault diagnosis. The proposed method is based on the joint application of morphological component analysis (MCA) and tunable Q-factor wavelet transform (TQWT) to decompose a signal into two signal components (i.e., low- and high-oscillation components) according to whether they having sustained oscillations. As bearing fault-induced transients (low-oscillation component) oscillate differently from periodic interferences and noise (high-oscillation component and residual), they can be separated via the MCA with the aid of TQWT which is parameterized by Q-factor and plays a role of distinguishing signal components presenting different oscillatory behaviors. The low- and high-oscillation components can be obtained by solving the objective function formulated based on MCA and TQWT. The determination of Q-factor for each signal component representation is also explored in this paper. The effectiveness of the proposed method is examined by experimental data.