Small sample bearing fault diagnosis based on S-transform combined with CNN-SVM

In response to the problems of long training time and poor classification performance of Convolutional Neural Network (CNN) in small sample bearing fault diagnosis, this paper proposes a diagnostic model based on S-transform and CNN-SVM. Firstly, the bearing vibration acceleration signal is transformed into a time-frequency image through S-transform (ST), and CNN is used to extract features from the ST time-frequency image to obtain deep features. Then, after regularizing these deep features, they are input into Support Vector Machine (SVM) for feature classification, fully utilizing the model's classification ability and improving the accuracy of bearing fault classification. The experiment shows that this method still has good diagnostic performance under small sample signals and noisy conditions. Compared with other methods, it has better noise resistance stability and stronger generalization ability.