Abstract: In order to realize the intelligent fault diagnosis and identification on the bearing of oil pump, an intelligent fault diagnosis method for the roll bearing of oil pump was proposed based on the feature of abundant fault information in the S transform contour line time-frequency map of vibration signal of the bearing of oil pump. In this method, the feature extraction is conducted on the S transform contour line time-frequency map by means of texture analysis. By virtue of this method, the vibration signal of bearings is S transformed and the contour line time-frequency map obtained on the basis of S transform is analyzed as the texture image. The roughness, contrast and directionality of its Taruma texture feature are extracted to act as characteristic vectors and the support vector machine is used as the classifier to implement the intelligent fault diagnosis on bearings. The practical test results of bearing fault signals indicate that this method can provide highprecision fault classification, identify the roll bearing fault automatically, so it is of hign value in engineering application.