Rolling bearing fault feature extraction method based on GWO-optimized adaptive stochastic resonance signal processing

Abstract

The failure of rolling bearings affects the function and precision of rotating machinery significantly, which has drawn lots of attention in this field. Dealing with the failure of rolling bearings, fault feature extraction is the first and most important problem. In this work, we convert the bearing fault signal into stochastic resonance dynamics equivalently. And, adaptive stochastic resonance is adopted to extract the fault signal feature. In addition, for industrial application of fault signal processing with large amplitude and noise intensity greater than 1, normalized scale transformation is introduced into adaptive stochastic resonance and then solved by fifth-order Runge–Kutta algorithm. Then, to further optimize the solving precision of stochastic resonance model, the scaling coefficient and step size of Runge–Kutta algorithm are chosen with the help of Grey Wolf Optimizer (GWO). Thus, we can obtain a fast convergence speed, high calculation accuracy and effective improvement of signal-to-noise ratio fault feature extraction method for rolling bearing fault signal processing. Finally, a comparation simulation was carried out to demonstrate the efficiency of the proposed method. Compared with Cuckoo Search Optimizer-based stochastic resonance signal processing method, the proposed method achieved a higher signal-to-noise ratio (SNR) to benefit the fault feature extraction. In summary, this work gives out a more practical and effective solution for rolling bearing fault feature extraction in rotating machinery fault diagnosis field.