论文信息 - Research on Fault Diagnosis Algorithm Based on Structure Optimization for Convolutional Neural Network

Research on Fault Diagnosis Algorithm Based on Structure Optimization for Convolutional Neural Network

Most of the traditional fault diagnosis methods rely on artificial extraction features and the expert knowledge of related fields, and these algorithms are not accurate, and the ability of robustness and generalization are poor. Convolutional neural network is one of the most widely used deep learning models. Based on its unique convolution-pooling network structure, convolutional neural network has powerful feature extraction and expression capabilities. In this paper, the structure of classical convolutional neural network is optimized, and a fault diagnosis algorithm based on structure optimization for convolutional neural network is proposed. For the characteristics of one-dimensional vibration signals, the convolution kernel of the first convolutional layer is set to a wide volume, and we use the batch normalization algorithm to further improve the convergence speed of the model. Through the experiment of the CWRU bearing fault public data set, the proposed algorithm has more than 99% fault recognition rate.

[1] ZhiQiang Chen,et al. Gearbox Fault Identification and Classification with Convolutional Neural Networks , 2015 .

[2] Sergey Ioffe,et al. Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning , 2016, AAAI.

[3] Lei Hang,et al. Survey of convolutional neural network , 2016 .

[4] Cong Wang,et al. Construction of hierarchical diagnosis network based on deep learning and its application in the fault pattern recognition of rolling element bearings , 2016 .

[5] Huaqing Wang,et al. Intelligent diagnosis method for rolling element bearing faults using possibility theory and neural network , 2011, Comput. Ind. Eng..

[6] Bing Li,et al. A new multiscale noise tuning stochastic resonance for enhanced fault diagnosis in wind turbine drivetrains , 2016 .

[7] lt,et al. Fault diagnosis method based on sensitive feature selection and manifold learning dimension reduction , 2014 .

[8] Chen Lu,et al. Fault diagnosis of rotary machinery components using a stacked denoising autoencoder-based health state identification , 2017, Signal Process..

[9] Jimmy Ba,et al. Adam: A Method for Stochastic Optimization , 2014, ICLR.

[10] Ming Zhao,et al. A convolutional neural network based feature learning and fault diagnosis method for the condition monitoring of gearbox , 2017 .