Rolling Bearing Fault Diagnosis Based On Stochastic Resonance And EMD

As one of the important rotating machines,rolling bearings have a very important influence on the operating state of the entire mechanical device.However,due to the harsh working environment,the fault characteristic information of the rolling bearing is often submerged in the noise and difficult to extract.Although many signal processing methods have achieved good application effects in the field of bearing fault feature extraction and diagnosis,there are still many problems.For example,the EMD method sometimes exist modal mixing problem and it is fail to decompose the characteristical signal successfully.Moreover,when studying the stochastic resonance method,the noise mostly focused on Gaussian white noise,but the actual engineering contains various kinds of noises.In addition,the signal-to-noise ratio usually used as the evaluation index in the stochastic resonance method,but this indicator needs to know the accurate fault frequency in advance.These problems will directly affect the extraction of weak feature signals,which will make the bearing fault diagnosis become more difficult.Aiming at the above problems,this paper takes rolling bearing as the research object,and uses stochastic resonance and EMD method to research the fault feature extraction and fault diagnosis of rolling bearings under strong noise background.The specific contents are as follows:(1)Rolling bearing fault diagnosis based on the combination of stochastic resonance and EMD under white noise background.When decompose the multi-frequency signals by EMD method,sometimes it is fail to decompose signals successfully and exhibit modal mixing problem.Therefore,this paper proposes a method based on the combination of adaptive stochastic resonance in periodic potential system and EMD.Meanwhile,the method which combing adaptive stochastic resonance in bistable system and EMD is compared with the method which combing adaptive stochastic resonance in periodic potential system and EMD.The results of simulation and experimental signals show that the combination of adaptive stochastic resonance in bistable system and EMD is also prone to appear modal mixing problem,but the combination of adaptive stochastic resonance in periodic potential system and EMD can decompose the characteristic signal successfully.(2)Low speed rolling bearing fault diagnosis based on the stochastic resonance in bounded noise background.Bounded noise is relatively common in practical engineering.Herein,this paper uses the stochastic resonance method to diagnose the low-speed rolling bearings fault under bounded noise background.We specifically study three cases,where the bounded noise frequency is much larger than the signal frequency,the bounded noise frequency is greater and close to the signal frequency and the bounded noise frequency is close and less than the signal frequency.The results show that the stochastic resonance effect is better when the bounded noise frequency is much larger than the signal frequency.When the bounded noise frequency is greater and close to the signal frequency,fault frequency is still identified but contains more interference components.When the bounded noise frequency is close and less than the signal frequency,the effect of stochastic resonance is unsatisfactory,and it is necessary to introduce vibration resonance and auxiliary signals to help extract fault characteristic information.(3)Rolling bearing fault diagnosis based on the stochastic resonance with improved signal-to-noise ratio.In stochastic resonance,the classical evaluation index is the signal-to-noise ratio.However,calculating the classical signal-to-noise ratio requires knowing the accurate fault frequency in advance.In engineering practice,the detected precise fault frequency is unknown in advance.Therefore,this article proposes the improved signal-to-noise ratio index.This new indicator does not need to know the exact characteristic frequency in advance.Obtain the fault theoretical frequency by calculation formula according to the calculation formula of the corresponding type of rolling bearing and then search the actual exact fault frequency near the fault theoretical frequency.By comparing the stochastic resonance based on classical signal-to-noise ratio with the stochastic resonance method based on improved signal-to-noise ratio,the simulation signal and experimental signal prove the feasibility and effectiveness of this proposed method.(4)Rolling bearing fault intelligent diagnosis system based on Lab VIEW and MATLAB.According to the bearing fault diagnosis method based on stochastic resonance with improved signal-to-noise ratio from previous chapter,Lab VIEW and MATLAB programme the rolling bearing fault intelligent diagnosis system.In the interface,by inputing the bearing type,bearing structure parameters and speed,then the rolling bearing intelligent diagnosis system based on the stochastic resonance with improved signal-to-noise ratio starts running.Finally,fault feature and fault type of rolling bearing can be extracted and discriminate.