Vibration Characteristics Analysis Of Fault Bearing And Typical Fault Diagnosis

Rolling bearing is the most widely used part and liable to be damaged in a variety ofrotating machinery, whose running state often directly affects the performance of thewhole machine. Diagnosing the fault of rolling bearing can find the faults timely and canprotect the property and workers in factory. Especially in steel rolling machinery, rollingbearing is particularly important part. Thus, researching fault diagnosis of rolling bearinghas great practical significance.This thesis makes use of electric discharging machining technique to breakup rollingbearing to simulate fault of rolling bearing. Then this thesis builds experimental bench ofrolling bearing and picks up vibration signal of rolling bearing which is used forexperimental analysis on experimental bench of rolling bearing.Firstly, this paper analyses the failure mechanism and vibration characteristic ofrolling bearing, the vibration signal characteristics of the rolling bearing with localdamage are summarized as cyclical shock and amplitude modulation. Then, in order toeffectively extract fault signature of rolling bearing, a fault diagnosis study based on EMDand FastICA combined with is put forward. Through processing and analyzing thepractical rolling bearing experiment data, it is showed that this method can accuratelyextract the fault characteristic of rolling bearing. In the next place, the theory of supportvector machine is studied and the genetic algorithm based on cross-validation is putforward to find the optimal parameter C and γ meanwhile to ensure the optimality ofthe classification model. Finally, a method is suggested to use the singular value ofseparation matrix as rolling bearing fault feature vector. Through doing the actual rollingbearing vibration experiment and analyzing the practical rolling bearing experiment data,it is proved this way can better distinguish fault type exactly.The intelligent fault diagnosis system platform of rolling bearing based on MATLABand LabVIEW is developed. The platform has friendly interface and is easy to handle.Several typical faults are simulated on this platform, the result showed that this systemplatform does well.