Research on Intelligent Fault Diagnosis for Rolling Bearing Based on Support Vector Machine
|School||Shenyang University of Technology|
|Keywords||rolling bearing fault diagnosis support vector machine|
Rolling bearing is one of the most ordinary parts in mechanical machines, and thefrequency of fault is high. Its working condition influences on performance of the wholemachine, even the whole production line. Therefore, it is very essential and important tostudy the fault diagnosis of rolling bearing.Rolling bearing is regarded as the research object, and the basic structure and failuremodes of the bearing are analyzed in this paper. Then the fault mechanism and vibrationfeature are studied systematically. Based on the characteristics of rolling bearing faultsignal, acceleration sensor is used to collect vibration signal from four different workingcondition of bearing: the normal, inner fault, outer fault, rolling body fault.Because the traditional fault diagnosis requires a lot of data samples, while the realtest data sample is not easy to obtain, it brings great difficulty to fault diagnosis.Therefore，according to the feature of the smaller number samples, the method of SVM isapplied to the fault diagnosis of rolling bearing, which offers a new research method forintelligent fault diagnosis. The research method based on theoretical research andcomputer simulation is proposed. First the vibration signal is collected by the sensor, thenthe noise of vibration signal is removed by the method of wavelet threshold. Theexperiment data is transformed by wavelet packet, the vibration signal is decomposedinto the individual frequency bands. The energy spectrum feature vectors are extractedfrom the individual frequency bands, and they are set as the input vectors of SVM.Finally, the running condition of the rolling bearing is diagnosed intelligently by theanalysis method of SVM.The computer analysis of this paper is all based on the software of MATLAB. Atthe end, the experimental results show that, the proposed method can diagnose the faultof rolling bearing with smaller number samples more accurately.