Bearing fault diagnosis method based on information fusion and fast ICA

Here, aiming at the problem of signals monitored by vibration sensors being easy to be interfered by noise, a bearing fault diagnosis method based on fast ICA algorithm and information fusion was proposed. Firstly, this method used the fast ICA algorithm to de-noise signals measured at each channel. Then an adaptive linear weighted fusion algorithm was used to perform data layer information fusion for the de-noised signals. Finally, an adaptive band-pass filter was designed based on the spectral kurtosis index to extract features. This method solved bearing fault feature extraction problems under the condition of low SNR. Simulated and actual test bearing fault signals were used to verify the effectiveness of the proposed method. © 2020, Editorial Office of Journal of Vibration and Shock. All right reserved.