Research On Intelligent Fault Diagnosis Of Rolling Bearing Based On Non-stationary Signal Processing

The rolling bearing is one of the most widely used and damageable parts in rotating machinery with its working state directly impacting on the performance of the whole machine, and the fault diagnosis of rolling bearing has a very important practical significance.The fault diagnosis of rolling bearing based on vibration signal analysis is one of the most commonly used and effectively methods.However,the vibration signals of the running bearing are typical non-stationary random signals.The key step is how to availably extract the fault features that can accurately reflect the state of the bearing from the vibration signals.According to the problems of the conventional empirical mode decomposition(EMD), the determine methods of extrema and sifting stoppage criteria are improved.A new idea is presented, regarding the correlation coefficient with some separability of the original signal and the intrinsic mode function(IMF) are obtained by using EMD as the features of the bearing state.According to the problem that the stationary artoregressive model cann't generally and accurately describe the nonstationary vibration signals. After adopting wavelet-packet to decompose the nonstationary vibration signal into stationary signals,a new idea regarding the parameters of the autoregressive model of the node coefficient are obtained by using wavelet-packet decomposition as the bearing's runnig state characteristic is presented.After the time-vary autoregressive model of the vibration signal is established,a new method is proposed,regarding the means of parameters as the state characteristic.According to the problem that single characteristic doesn't have higher discrimination for fault state of differnt types and degree under different running conditions and the random compounding of every type characteristic cann't effectively improve the discrimination too.An intelligent fault diagnosis method based on feature selection and support vector machine(SVM) is proposed.Firstly,the combined feature set is constructed by using the four feature extraction methods in this article,then the sensitive features subset using feature selection method based on distance evaluation technique is selected.Finally,the sensitive feature subset is sended into the SVM classifier to make faults classing and diagnosis.Large numbers of experiments show the proposed methods can effectively and accurately identify the fault state of deffernt type and degree under different running condition.