A fault diagnosis method for bogie axle box bearing based on sound-vibration multiple signal fusion

The fault diagnosis method of multi-signal fusion is one of the current research trends, which can improve the reliability of diagnosis results. In this paper, the single channel signal is decomposed by multi-channel bandpass filter bank, and a new indicator value is constructed to select the optimal component. A new fusion demodulation method is constructed by using the two signal demodulation methods to extract the characteristic frequency of the single channel signal. Subsequently, the characteristic spectrum of the multi-channel signals is fused to extract the final characteristic frequency. The diagnosis method is verified by the simulation signal and the sound signal and vibration signal collected by the experiment. The results show that the proposed method can reduce the content of noise components in the characteristic spectrum, highlight the fault characteristic frequency, and reflect the superiority of the proposed method compared with other methods. This paper provides guidance for feature extraction of data fusion methods in the future, and provides an effective method for fault diagnosis and condition monitoring of bearings.