Optimization of Bearing Fault Diagnosis Algorithm Based on Convolutional Neural Network

Traditional fault diagnosis methods have the problems of limited feature extraction and inaccurate fault detection. Rotor axis locus, as one of the typical features of rotor faults, can provide more representative fault feature information. Accurate recognition of rotor axis locus shape is the basis of constructing rotor fault characteristic signs. A bearing fault diagnosis method based on deep convolutional neural network is proposed to improve the generalization ability of rotor axis track shape recognition. The rotor axis locus is mapped to the two-dimensional digital image, and the extracted two-dimensional digital image is processed by the strong ACON activation function. The pre-processed digital image of the rotor axis locus is input into the deep convolutional neural network for training. The results show that the digital image of rotor axis locus extracted by strong ACON activation function processing is more advantageous in the training of deep convolutional neural networks. Compared with the traditional rotor axis path shape recognition method, the rotor axis path shape recognition method based on deep convolutional neural network has higher accuracy and better robustness.