Stochastic Resonance-based Weak Signal Detection For Fault Diagnosis Of Rolling Bearing

Mechanical breakdown or accidents will cause serious economic losses and also pose a threat to people's properties and lives.So it is crucial to diagnose the faults of mechanical equipment and ensure the safe and stable operation of equipment.Usually the vibration signal,sound signal and current signal of mechanical equipment are collected and then analyzed to evaluate the operating state of the equipment while the machine is working.However,the signals collected by sensors generally need to be analyzed on PC,the real-time performance is poor,and so the optimal time for mechanical fault diagnosis may be missed.Besides,the collected signals often contain serious running noise and environmental background noise,which makes signal analysis difficult and may cause misjudgment of the operation status of mechanical equipment.Therefore,it is particularly important to study fast fault diagnosis methods,suppress noise and extract feature information of weak signals in mechanical fault diagnosis.In this paper,a bearing fault diagnosis method based on embedded stochastic resonance is studied.Nowadays,Stochastic Resonance(SR)algorithm is widely used in fault diagnosis of rotating machinery.While,most SR algorithms are developed on desktop platforms,which focus on off-line signal analysis.But in this paper,it proposes a simple and easy-to-implement on-line bearing fault diagnosis method based on embedded system,which has implemented signal denoising and filtering and then process filtered signal by stochastic resonance.Regards signal-to-noise Ratio(SNR)as the criterion to identify the fault type of bearing and display the diagnosis results on LCD.Adaptive Stochastic Resonance(ASR)method has been proved that it can effectively enhance the weak periodic signal which has submerged in strong background noise.Also,the ASR system does not rely on SNR as the tuning index,so when the target fault frequency can't be estimated accurately,ASR plays a not bad effect on eliminating noise.However,when the SNR of vibration signal is extremely low,conventional ASR method may not correctly increase the fault characteristic frequency(FCF)energy.So in order to solve this problem,this paper presents a method of bearing fault diagnosis based on Sound-Aided Vibration Signal ASR(SAVASR).Firstly,demodulate the bearing sound and vibration signals.Enhance adaptively the envelope vibration signal by moving the sliding window along the time axis of the envelope acoustic signal.Then send the optimized fusion signal to the ASR system,the parameters will be adjusted adaptively under the guidance of the comprehensive evaluation index.Finally,use the spectrum of the optimal SASASR output signal to detect the bearing fault.By qualitative and quantitative analysis,the performance of the proposed SAVASR method is fully evaluated comparing to the traditional ASR method which only processes vibration signals.In summary,this paper studies the bearing fault diagnosis method based on embedded SR and sound signal auxiliary vibration signal.Meanwhile,a multi-sensor information fusion method is also proposed,which can enhance the weak signal of vibration by using the energy of sound under the condition of low signal-to-noise ratio of vibration signal,so that realize the enhancement of the characteristic frequency of bearing weak fault under strong background noise.Compared with the conventional method,this method is more convenient and easy to implement,and also improves the accuracy of bearing fault diagnosis.