Research On Bearing Fault Diagnosis Method Based On Deep Learning

In the development process of modern industry,industrial equipment gradually tends to be complex,precise and intelligent,and once the equipment fails,it will lead to the paralysis of the whole industrial process.Therefore,how to quickly and accurately diagnose the faults of equipment has become an important problem that can not be ignored in the field of Industrial Internet.Taking the bearing as the research object,this paper models the different problems encountered in the process of bearing fault diagnosis,puts forward the corresponding fault diagnosis model,and verifies the effectiveness of the model through experiments.The research contents of this thesis are as follows:(1)Aiming at the problem that the traditional fault diagnosis methods rely too much on manual experience and can not completely extract features,a fixed working condition bearing fault diagnosis method based on dual channel is proposed.This method fully extracts the features contained in the signal through the dual channel model,the features are weighted and fused by the improved Squeezeand-Excitation Networks,and finally the fused features are classified by Softmax classifier.Experiments show that this method has good fault diagnosis performance,which proves the effectiveness of the proposed dual channel feature fusion model and the improved Squeeze-andExcitation Networks.(2)Aiming at the problem of insufficient data set caused by working condition change in actual industrial production,a variable working condition bearing fault diagnosis method based on Transfer Learning is proposed.Firstly,the source domain sample set is input into the model for training,and the features of source domain and target domain are extracted respectively;Then,Multi-layer and Multi-kernal Maximum Mean Difference method is proposed to quantify the difference between the features extracted from the two domains,and a new loss function is designed to fine tune the model;Finally,the migrated model is used for bearing fault diagnosis.Experiments show that the proposed method can effectively solve the problem of insufficient data under variable working conditions.