A compound fault diagnosis method of rolling bearing based on wavelet scattering transform and improved soft threshold denoising algorithm

Abstract

The vibration signal of faulty rolling bearing of rotating machine carries a large amount of information reflecting its fault categories. However, compound fault features are easily mixed together, and can cause missed diagnosis and misjudgment, which is still a challenging task in mechanical fault diagnosis. A compound fault detection method using wavelet scattering transform (WST) and an improved soft threshold denoising algorithm is proposed to extract compound faults in bearings. First, the wavelet scattering transform is used to calculate the original scattering coefficients from vibration signals. Second, the improved soft threshold denoising algorithm is applied to obtain the renewable scattering coefficients, which are further employed to reconstruct the denoising signals. Third, process the envelope spectrum analysis on the denoising signal to extract fault features. Finally, both the simulations and experiments in associate with comparison investigations proved that this method can effectively detect compound faults in bearings.