Research On Fault Modeling And Intelligent Diagnosis Method Of Rolling Bearings Based On Digital Twin

Due to the complex and changeable operating environment of rolling bearings,there are differences in the data distribution of the same fault category under different working conditions,and the labeled data under some working conditions are not easy to be obtained,and it is impossible to build an effective model.Research on rolling bearing fault diagnosis technology has practical significance.In view of the above problems,it is proposed to use the digital twin technology as the core,combined with the deep learning method to construct the rolling bearing fault diagnosis model,and use a small amount of labeled data to realize the rolling bearing fault diagnosis task under different working conditions.First of all,in view of the difficulty in obtaining the training data labels of rolling bearings and the small number of samples under different working conditions,it is proposed to realize the digital twin modeling of rolling bearings by using finite element analysis software,and establish the rolling bearing fault twin models under different working conditions.The central difference algorithm of formula dynamics is used to simulate it to obtain rich twin data.After time-domain signal comparison and envelope spectrum technical analysis,it is verified that the twin data samples are highly similar to the real bearing data samples.Secondly,the generation confrontation network based on Wasserstein distance is adopted,and the gradient penalty item is introduced to improve it,reduce the distribution difference between twin data and real data,realize feature fusion,and propose a combination of digital twin technology and improved Wasserstein GAN algorithm is used to fuse twin data and real data to reduce the distribution difference between them.According to the analysis of experimental results,using feature fusion method can significantly improve the accuracy of rolling bearing fault diagnosis task under the same working conditions,which not only ensures the authenticity and reliability of twin data,but also lays the foundation for the next step of rolling bearing fault diagnosis tasks under different working conditions.Finally,using the idea of transfer learning,the global attention mechanism is introduced to improve the residual network,and the multi-core maximum mean difference is used to perform domain adaptation processing on the features extracted from the source domain and the target domain,so as to realize the feature transfer of the unlabeled target domain sample data,and finally establish an intelligent model for rolling bearing fault diagnosis based on digital twins.The proposed method can effectively solve the problem of missing labeled samples of bearings under different working conditions through experiments with 12 transfer tasks,and has a high diagnostic accuracy.At the same time,by setting up rolling bearing transfer tasks under different specifications for comparative experiments,it is proved that the proposed method has good generalization.