Abstract:
Need for traditional method of rolling bearing fault diagnosis algorithm and select the feature extracting in construction and shallow model to dig the mapping relations between the signal characteristics and equipment of health problems, put forward the application of convolution neural network (Convolutional Neural Networks, CNN) and gray Wolf algorithm (Grey Wolf Optimizer, GWO) optimized combination of support vector machine (Support Vector Machine, SVM), a new method of fault diagnosis. First converting the original signal to measure spectrum diagram, and then select the trained CNN model Alexnet for feature extraction, the scale of the signal spectrum and characteristic is obtained by principal component analysis (Principal Component Analysis, PCA) for data dimension reduction, the final will be normalized data input to the wolves algorithm (GWO) after the optimized SVM classifier, so as to realize the fault diagnosis of bearing health status. The simulation experiment with bearing data provided by western reserve university shows that the proposed method can extract appropriate features adaptively and has a high classification accuracy.