Research On Bearing Fault Diagnosis Method Based On Deep Learning Of Attention Mechanism

Over the past few years,deep learning-based techniques have been extensively and successfully adopted in the field of fault diagnosis.Convolutional neural network(CNN)and recurrent neural network(RNN)are popular in end-to-end deep learning networks.As the diagnosis tasks become more complicated,the structures of the traditional deep learning methods have to become deeper to deal with them,while the gradient of fault features may vanish within the deep network.In addition,all the features are treated equally in the traditional models,which cannot make the most of the feature extraction capabilities of CNN and the ability to mine time series information of RNN.In order to effectively diagnose faulty bearings from high-noise real working conditions,a method named dual attention dense convolutional network(DADN)is proposed to classify faulty bearings end-to-end.It is constructed by the dense network and the dual attention block.On one hand,the dense connections and concatenation layers in DADN can reinforce the propagation of fault features among layers and mitigate the vanishing gradient phenomenon in the deep network.On the other hand,as the features flow through the channel attention and spatial attention within the dual attention block,this attention mechanism can learn which feature to emphasize or suppress and then obtain the cross-channel and cross-spatial weights of the features.These weights can make the most of the abundant information,elevating the expressive power of network.In order to combine the feature extraction ability of DADN with the long-term effective "memory" ability of LSTM for time series information,an attention dense-LSTM network(ADLN)model that combines DADN,LSTM and another attention module is also proposed to predict the remaining useful life of the bearings.DADN extracts high-dimensional features from the original diagnostic signals,and then the LSTM network extracts the time series information containing the bearing remaining life,and finally gives the predicted value of the bearing life.In order to verify the performance and generalization ability of the above models,a comprehensive comparison with other common models was carried out on multiple data sets.The final results show that the DADN and ADLN models proposed in this paper have greater advantages over existing models in terms performance,robustness and generalization capabilities.