A fault diagnosis method based on improved convolutional neural network was proposed to address issues such as delayed manual diagnosis of abnormal information of business center resources in customer service and low fault identification rate.After the convolutional layer,a batch normalization layer was introduced to improve the generalization ability of the model.The Drupout function was introduced in the fully connected layer to alleviate overfitting problems.The data was also enhanced and the grey wolf algorithm was used to optimize hyperparameters.The model was simulated in Python and Pycharm environments.It is found that the test set accuracy of the classical convolutional neural network is around 85％,while the improved test set achieves a result of around 94％,indicating that the proposed design has significant effects.
关键词
卷积神经网络/批量归一化/Dropout层/灰狼算法/台资源
Key words
convolutional neural network/batch normalization/Dropout layer/grey wolf algorithm/business center resources
维普
万方数据