Fault Diagnosis of Fan Bearing Based on Improved Convolution Neural Network

Because of its accuracy, Convolutional Neural Network (CNN) has become an important method in the field of fault diagnosis. However, the traditional CNN has a long time of training and diagnosis due to its complex structure. At the same time, due to many problems in the network, the detection accuracy is not high. Therefore, this paper proposes an improved CNN for fan bearing fault diagnosis, which speeds up the feature extraction of the network by improving the network structure; solves the problem of part of neurons not being activated by improving the activation function, and improves the accuracy of network detection. Finally, the network proposed in this paper is validated on the data set and compared with other advanced fault diagnosis algorithms. The results show that the accuracy of the algorithm proposed in this paper can reach 99.76%. Because of other algorithms, and the training and diagnosis time is relatively short, it has practical application value.t).