Aiming at the problem that the current unsupervised cross-domain fault diagnosis methods are often difficult to align the edge distribution and the conditional distribution at the same time

which leads to low diagnosis accuracy

a joint distributed adaptive adversarial network fault diagnosis method is proposed. Firstly

the original vibration signal is preprocessed by using fast Fourier transform

and the residual neural network is constructed to extract the deep features of the samples. Secondly

the class discrimination information passed by the classifier is used to assist the conditional domain adversarial network to reduce the distribution difference between domains. At the same time

the combined maximum mean difference and the output information of multiple fully connected layers are introduced to realize the adaptive edge distribution and conditional distribution of the cross-domain diagnostic model. Finally

the proposed method is tested by using the data set of Paderborn University in Germany and the data set of laboratory bearings

and the effectiveness and feasibility of the proposed fault diagnosis method in different migration scenarios are proved.