Bearing Malfunction Diagnosing Research And Application Based On EMD And ANN

Roller bearings are one of most used and important mechanical parts in industrial production, and are vulnerable to damage. Many faults of rotating mechanism are related to the state of roller bearings. The performance of roller bearings directly affects the performance of the whole machine. Thus, developing state testing and fault diagnosis system of roller bearings is very necessary.Based on widely reading of references and understanding of bearing vibration test technology and bearing fault analysis theory, this thesis analyzes the resonance demodulation method in detail and puts forward the adaptive linear enhancer(ALE) which can decrease the noise and improve the spectrum of the bearing signals. Considering the bearing signals are non-stationary, this thesis introduces a powerful tool for analyzing the non-linear and the non-stationary signals which is called empirical mode decomposition(EMD). Compared with the traditional resonance demodulation method, EMD can exactly locate the source of fault. Through a large number of experimentations, the relation between sampling number and distinguishing accuracy is obtained. Finally the neural network is improved to distinguish different faults by using many kinds of eigenvalues.