Research on Rolling Bearing Fault Diagnosis Method of Fuzzy Broad Learning System Based on Genetic Algorithm Optimization

The most common motor fault is the failure of bearing, and the traditional fault diagnosis method requires manual extraction of features, which has high uncertainty and complexity. In this paper, a motor bearing fault diagnosis method based on Gramian angular field image coding and CNN-SVM is proposed, which has three stages, firstly, the original bearing vibration signal is converted into a two-dimensional scale diagram by using Gramian angular field image coding. Second, next, an enhanced CNN model is employed to quickly and accurately identify the elements of the converted image. Finally, SVM are used as final classifiers to further improve the accuracy and speed of fault classification. In the experimental design, the model was verified and analyzed several times by using the bearing fault dataset of Case Western Reserve University, which verified the feasibility and superiority of the method and provided a theoretical basis for guiding bearing maintenance decisions.