Bearing fault feature extraction using autoregressive coefficients, linear discriminant analysis and support vector machine under variable operating conditions

Abstract

Advanced monitoring requires automatic diagnosis of machines operating under variable conditions. In this paper, an intelligent method is introduced in order to enhance the classification and achieves a higher precision for the diagnosis of degradation of rolling bearings operating under condition variations. The method uses the coefficients of autoregressive modeling (AR) of the bearing vibration signal as the features of a classifier. A Linear Discriminant Analysis (LDA) of the matrix feature obtained from AR analysis is applied in order to extract the components that discriminate the different fault modes since it is insensitive to the working conditions. Finally, the results obtained from LDA are used as the input of a support Vector Machine (SVM) classifier to automatically identify the bearing state. The experimental results show that the performance of the proposed method is effective and achieve a good accuracy.