Rolling Bearing Incipient Fault Detection via Optimized VMD Using Mode Mutual Information

The complete failure of the rolling bearing is a deterioration process from the incipient weak fault to the severe fault, thus it is important to alarm when the incipient fault appear. This work presents a novel incipient bearing fault diagnosis framework using mode mutual information (MMI) based fitness function, variational mode decomposition (VMD), and cuckoo search (CS) algorithm. MMI based fitness function is proposed in order to obtain the optimal combinations of the VMD parameters. Therefore, the optimal parameters of VMD can be obtained by CS algorithm using proposed fitness function. Afterwards, a vibration signal is decomposed into a set of modes using the optimal VMD, and the kurtosis value of all modes are calculated. The envelop of the mode with maximum kurtosis value between modes and raw signal is computed as the input vector of the stacked denoised autoencoder (SDAE). Comparisons have been conducted via SDAE to evaluate the performance by using EMD and the fixed-parameter VMD. The experimental results demonstrate that the proposed method is more effective in extracting the incipient bearing fault characteristics.

Automated summary

This study proposes a novel framework for incipient bearing fault diagnosis using MMI, VMD, and CS algorithm, achieving more effective fault characteristic extraction through optimizing VMD parameters and feature extraction.