The Research Of Fault Diagnosis Methods Of Ball Bearing Based On The Wavelet Analysis

The roll bearing is the most important part of mechanical equipment, which hasthe merits of low friction resistance, installing convenience and lubricate easilyrealized. So it is commonly used in rotated machine. The stand or fall of bearingaffects straightly the mechanical working state, any fault or invalidation took place inthe running of the machine will bring serious sequence and great economic loss. So itis necessary to check the state of the bearing and diagnose the fault.In chapter 1, the intention and meaning of the paper is introduced briefly. Thedevelopment of fault diagnoses technology, the fault diagnoses of roll bearingresearch actuality with Wavelet transform, the research of intelligent fault diagnosesbased on neural network, and the research content of the paper are introduced.In chapter 2, the basic invalidation form, vibration form, vibration mechanismand vibration diagnose methods of the roll bearing are discussed in detail.In chapter 3, FFT and Windowed Fourier transform which describe thefrequency information of the signal is introduced briefly. The wavelet transform isstudied chiefly, the fault signal of the roll bearing is analyzed with multi-scaleanalysis and the analytic result can be obtained.In chapter 4, the Wavelet packet is studied mainly, which divided thenot-subdivision of the high frequency further and the detail information can beobtained. The fault signal of the roll bearing is analyzed with multi-scale Waveletpacket and emulation with MATLAB.In chapter 5, the merits of time-frequency location of the Wavelet transform iscombined with the ability of self-study and nonlinear mapped of the NN, energycharacteristic of the fault is extracted, which constructs eigenvector, and the faults canbe classified by the RBF neural network. At the same time, the faults are classified byFisher linear criterion, which is compared with RBF.In the paper, the modern theory analysis methods of the wavelet and the artificialneural network are combined with the productive practice, which are used in the fault diagnosis of the ball bearing. It is a useful attempt, and better result is obtained.