Study On Rolling Element Bearing Fault Diagnosis Methods Based On Wavelet Theory

Mechanical fault diagnosis has always been a hot research and some achievementshave also been made. However, there are still a lot of problems in this field. The rollingbearing is one of the most commonly used components. Fault diagnosis to rollingbearing is of great significance.The diagnosis method in this paper is based on the vibration signal. The vibrationsignal of rolling bearing contains fault characteristics and is suitable for various bearing.Fault diagnosis based on the vibration signal is an effective diagnostic to earlier tinyfault, and the result is accurate and reliable.In this paper, based on the research of Wavelet theory, fault feature extraction is bythe method combining wavelet packet analysis and the logarithmic energy entropy.Fault feature extraction is particularly important in rolling bearing fault diagnosis. It isthe basis of bearing state recognition, and directly related to the accuracy of thediagnostic result. Firstly, the vibration signal is decomposed by wavelet packet todifferent frequency band signals. Then, in order to extract the fault feature, normalizedenergy entropy of each decomposed signals is calculated.In this paper, the least squares support vector machine is used to identify thebearing states, and the diagnosis result is quite ideal. The least squares support vectormachine (LSSVM) uses least square linear system as the loss function instead of thetwo planning method which is used by the traditional support vector machines,simplifies the complexity of computation, and improve the computational speed. In thispaper, based on the research of LSSVM theory, the LSSVM is applied.In this paper, an experimental platform of rotating machinery fault diagnosis isbuild. A variety of rolling bearing fault are simulated to test the fault diagnosis methodbased on wavelet packet energy entropy and LSSVM. In the end, the validity of themethod is proved. And the wavelet analysis algorithm is used to process rolling bearingvibration signal, given a preliminary diagnosis.