A fault diagnosis method based on S-transform and BP neural network is proposed according to the characteristics of bearing fault signal, such as unsteady state, strong noise and weak characteristics. Firstly, S-transform is used to transform the vibration signal into the time-frequency domain. The semi-soft threshold function is used to reduce the noise interference in the time-frequency domain. Then, the time-frequency domain analysis of the vibration signal after noise reduction is carried out. Four types of vibration signal features, including normal state, rolling body fault, inner circle fault and outer circle fault, are extracted. Finally, Grey wolf optimizer is combined with BP neural network to obtain the machine learning algorithm model after parameter optimization. The fault classification of rolling bearing is realized by identifying the vibration signal features. Compared with traditional methods, the proposed method has stronger anti-interference ability and higher diagnostic accuracy. Numerical simulation and open source experimental data further verify the feasibility and effectiveness of this method.

Fault diagnosis; S-transform; BP neural network; Vibration signal; Grey wolf optimizer