Research On Vibration Test And Intelligent Fault Diagnosis System Of Rolling Element Bearing

Rolling element bearing is one of the most widely used vital components in almost all kinds of machinery, and it is vulnerable to damage. Its quality and working condition has influence on performance of the machinery. So it is very significant to research on bearing vibration test and fault diagnosis techniques.Currently, vibration analysis method is the most common in bearing vibration test and fault diagnosis. Vibration signal of rolling element bearing is collected, and then its feature is extracted through signal processing. The running conditions of bearing are recognized by using methods of pattern recognition. Feature extraction and condition identification are pivotal. In this thesis, the methods of feature extraction from bearing vibration and the SVM used in bearing intelligent fault diagnosis system are studied. The outline of the work is as follows:The forming mechanism and failure mode of bearing vibration are introduced. The features of bearing vibration are analyzed.The statistical indexes of bearing vibration signal in time domain and frequency domain are analyzed. The feature of them when the bearing works in different conditions is also discussed.Principle and methods of zoom spectrum analysis are introduced. The application and experimental study of multiple modulation zoom spectrum analysis based on complex analytic band-filter is researched.The application of wavelet analysis and wavelet packet is studied. Threshold de-noising in wavelet domain is used for bearing vibration signal. Energy distribution at different frequency bands is obtained by wavelet packet decomposition. Energy distribution feature in frequency bands can be used as the feature sets of bearing fault diagnosis.The theory and application of support vector machine (SVM) is also researched. The statistical indexes of bearing vibration signal in time domain and energy distribution feature in frequency bands are used as input vector of SVM separately. In the thesis, two kernel functions, polynomial kernel function and RBF kernel function, are used in experimental analysis of SVM. The optimum structure and parameter of SVM used in bearing fault diagnosis are determined. It turned out that it can increase the diagnosis accuracy to choose proper kernel function and parameter of SVM. Finally, the settings module of bearing vibration measurement system and the intelligent fault diagnosis system is developed based on virtual instrument technology. In this thesis, SVM is applied to bearing fault diagnosis, which provides a kind of method to solve the intelligent diagnosis problem.