Fault Diagnosis Of Rolling Element Bearing Based On Wavelet Denoising And Empirical Mode Decomposition

Rolling bearing is the most widely used rotating part in mechanicalequipments, it is also the main fault source of the mechanical equipments, and itsrunning state always affects the performances of equipments directly. So thecondition monitoring and fault diagnosis of rolling bearing have importantpractical significance to avoid major accidents and huge economic losses causedby sudden failure and catastrophic destruction; it also has important practicalsignificance to maximize the potential and efficiency of equipment and topromote the economic development. However, the early fault bearing impactingenergy is very low, so the effect of background noise on the fault diagnosis oftencan not be ignored. The main research direction of rolling bearing fault diagnosisis to seek reasonable and effective methods of denoising and fault diagnosis.This paper designed a rolling bearing fault simulation test rig, it isconvenient and reasonable to change speed and load, and it also can rapidly andaccurately obtain the vibration signals of rolling bearing under differentconditions. Using the test rig, we carried out three kinds of simulationexperiments (normal rolling bearings, rolling bearings with outer race fault,rolling bearings with inner race fault, respectively), and collected the vibrationsignals of corresponding conditions.This paper used the improved wavelet denoising method for pretreat therolling bearing vibration signal, the main idea of this method is to improve thethreshold selection principle and the threshold function. Improved thresholdselection principle is on basis of characteristics on different scales of faultsignals and noise signals. Improved threshold function combines the advantagesof soft, hard function. The results show that the method can remove noiseseffectively, retain the fault signal components, so it provides a strong guaranteefor the feature extraction of fault signals.This paper combined the empirical mode decomposition (EMD) with theHilbert transform, and applies the method to extract the rolling bearing faultdiagnosis feature. After the test verification, this method can effectively extractthe rolling bearing fault characteristic frequency, and complete fault diagnosis.