Characteristic Signal Detection Based on Re-Scaling Frequency Stochastic Resonance and Its Application in Fault Diagnosis

Abstract

Stochastic resonance (SR) has been considered as a useful means for signal processing. However, under the condition of large parameters, for example, frequency and/or amplitude and/or noise level can be much larger than one, it is impossible to detect a weak characteristic signal submerged in strong noise or in complicated interferential components through SR. This is because SR is generally fit for small parameter signals. To realize the large parameter SR (SR-like), a new approach of re-scaling frequency stochastic resonance (RFSR) is proposed by means of a nonlinear bistable system, and hence the weak characteristic signal can be extracted. The RFSR can be considered as the observation of the SR phenomenon in different frequency-scale fields through the map of frequency-scale ratio. For non-SR status, the parameter property of two approaches of creating SR by adjusting system parameter and re-scaling frequency are investigated comparatively. The result of the investigation presents that, for large noise level, the two methods can both realize the recovery of SR status; but for large signal frequency, only the approach of re-scaling frequency can be used to recover SR status. However, if both approaches are combined to use, we can obtain a better SR effect. The experimental study of fault diagnosis for rolling bearing further demonstrates the effectiveness of the bistable RFSR approach.