A two-stage sound-vibration signal fusion method for weak fault detection in rolling bearing systems

Sound-vibration signal fusion methods are widely applied in fault diagnosis, but the acquisition of the sound signal is obviously affected by the position of the measurement points, and it is difficult to detect the weak fault characteristics under by strong background noise. In this paper, a two stage sound-vibration signal fusion algorithm is proposed, which enriches the states information of the bearing system and reduces the influence of background noise. In the first stage, the fault features of sound signals at multiple measuring points are combined and weighted by gray B type correlation degree, and then the features of signals are extracted by empirical mode decomposition and kurtosis superposition; In the second stage, the sound fusion signal and vibration signal are fused again by sampling frequency unification, and the weak fault detection of rolling bearings are realized by combining the fault characteristics of sound and vibration signals. Experimental results show that the two-stage signal fusion improves the fault feature detection accuracy significantly, and the signal-to-noise ratios of the fault features are enhanced obviously. This research provides a new fusion method for fault diagnosis of rolling bearings, which is helpful for the status monitoring of bearing systems.

Automated summary

In this paper, a two-stage sound-vibration signal fusion algorithm is proposed to enhance fault diagnosis of rolling bearings. The algorithm combines fault features from multiple measuring points and reduces background noise interference.