Bearing Fault Diagnosis Based on Scale-transformation Stochastic Resonance

A weak fault feature extraction method of rolling bearing based on scale-transformation stochastic resonance (STSR) is proposed. Combined with ensemble empirical mode decomposition (EEMD), the vibration signal with noise is adaptively decomposed for antialiasing by EEMD method to get intrinsic mode functions (IMFs) of different frequency bands, then the IMFs are inputted into scale-transformation mono-stable system. The low frequency fault features are extracted by using a frequency scale R to change the step length of numerical calculation and the adjustment of mono-stable system parameters, and finally slice bi-spectrum is adopted to perform the postprocessing of the output of the mono-stable system. Simulation analysis is performed to validate the characteristics of STSR, and analysis of measured signal of the rolling bearing with strong background noise shows that the approach can extract the weak fault features of rolling bearing successfully.