Bearing Fault Diagnosis Based on Improved 1D Convolutional Neural Network
In order to guarantee the secure and steady operation of mechanical equipment,this paper proposes a 1 D convolutional neural network fault diagnosis model with large convolution kernel and weak pooling structure.Firstly,the large convolutional kernel is designed to enhance the model's sensitivity to global features.Simultaneously,the pooling structure is simplified to further strengthen the model's abstraction capability for local features.Subsequently,batch normalization processing strategy is incorporated to achieve true recognition of fault situation.Finally,the model is confirmed by the Case Western Reserve University's bearing public dataset,and the results prove that the proposed model owns fantastic feature extraction skill and classification precision.
万方数据
维普
