Rolling Bearing Fault Diagnosis Based on Exact Moment Dynamics for Underdamped Periodic Potential Systems

This research is to build a more general bridge over the model investigation on stochastic resonance (SR) and the laboratory design in rolling bearing faults. To this end, we generalize the derivative matching moment closure method to the underdamped biased periodic potential systems to disclose the nonmonotonic evolution of the spectral amplification factor. With the exact moment dynamics available, a two-layer loop algorithm for detecting the incipient bearing faults is then developed. With the outer loop to optimize the output SNR for the best time scale factor and the inner loop to maximize the spectral amplification factor for optimal system parameters, the semi-analytic results are directly related to this laboratory application. The analog and experimental verification based on different datasets show that the proposed method can perform as well as the existing SR method, even under strong noise background. Particularly, the proposed method does not depend on the amplitude of the input signal when optimizing parameters, but only on noise intensity and characteristic fault frequency, thus it has higher detection efficiency than the existing simulation-based SR methods.