Research on Multiple Cross Correlation Fusion Method of Fault Vibration Signal

Auto-correlation and cross-correlation methods are combined with other signal processing methods, which are widely used in identifying fault sources and extracting fault features, but the extracted fault features are not obvious enough. Compared with the auto-correlation and cross-correlation methods in signal processing, this paper proposes a vibration signal multi-cross correlation fusion algorithm. This method associates the information obtained by the three vibration sensors to obtain a correlation coefficient. The Fourier transform simulation is used to compare and analyze the correlation coefficients of the pairwise cross-correlation and the direct multi-cross correlation of the fault simulation signal, highlighting the superiority of the direct multi-cross correlation method to obtain the signal three-correlation coefficient. Finally, the stacked auto encoder (SAE) diagnosis method is used to compare the fault diagnosis classification results of the original fault data and the multi-cross correlated fusion data in the test.