Fault feature enhancement for roller bearing using a sparsity-promoted method

Vibration signals induced by faulty roller bearing usually contain much interference, which increases the difficulty of fault diagnosis. Thus, it is significant to enhance the fault features and carry out noise reduction. To achieve fault feature enhancement for roller bearing, a novel method based on Majorization-Minimization (MM) algorithm is developed in this study. First, a sparsity optimization objective function is designed, which integrates impulsive feature preserving factor and penalty function factor. In this function, the regularization parameter is taken into consideration for corresponding different working situations. Second, a nonquadratic majorization iterative method based on MM is used to address the convex optimization problem of the designed function. A series of sparse coefficients can be got through iterating, which just contain the transient components. Finally, envelope analysis is used to extract the bearing fault features. Simulated signal and experimental data are applied to verify the effectiveness and applicability of the researched method. The results show that the bearing signals can be represented sparsely and the bearing fault features can be enhanced and extracted easily using the proposed approach.