Application Of Bearing Fault Diagnosis Technology In Digital Twin

In recent years,the progress of large data,artificial intelligence and the Internet of things in intelligent manufacturing technology have been more mature.Different from the previous mechanical production,the mechanical equipment has a large scale and the internal structure is more complex,the linkage between equipment is higher.As one of the transmission components with the highest frequency in mechanical equipment,because long-term load and wear are often the most prone to mechanical failures.Technical research on bearing fault diagnosis,classification,and performance degradation assessment can effectively avoid downtime caused by mechanical equipment failure.In this paper,bearings are taken as the fault mechanical subject,provides a new solution for bearing fault diagnosis and performance degradation evaluation by constructing a deep learning neural network and digital twin model.In the end,the experimental scheme is developed,and the proposed method’s actual effect is compared and verified.The main work is as follows:First,an innovative fault feature generation model is proposed to solve the unbalanced problem of the bearing fault data set.Extracting the existing fault feature,and fault category attributes by encoding and decoding then generates a new batch of faults for the network through the generation type.In order to verify the application of the new generation of fault features in actual fault diagnosis,a convolutional neural network was selected to classify the fault categories.Finally,through the comparison of the feature generation model and the comparative analysis of other feature generation models,it is proved that the fault diagnosis model’s accuracy,accuracy,and generalization ability has been improved.The proposed fault feature generation model can effectively improve the shortcomings of uncontrollability,simplicity and slow convergence speed of traditional feature generation models.Secondly,the performance degradation evaluation method can play the role of early warning and condition based maintenance.This paper designs and constructs the performance degradation feature extraction model and performance degradation evaluation model.First,the life cycle performance degradation characteristics of the data set are extracted,then adjust and update sensitive parameters in the trained offline network model.Moreover,the trained model is input into the complete life cycle data set,and the bearing performance degradation evaluation curve is drawn through online evaluation.Then,select the envelope spectrum analysis method to verify the performance degradation evaluation results at each stage.Compared with other methods,this paper puts forward the evaluation method of performance degradation can find fault in advance to provide the theoretical basis for the follow-up maintenance and trend prediction analysis.Finally,according to the theory of the model verification algorithm,the digital twin of fault diagnosis systems based on integrate industry and virtual reality is proposed.The aircraft engines are selected as mechanical equipment maintenance,built simulated sensor acquisition,equipment monitoring,data storage,fault diagnosis models,virtual cloud platform,and human-computer interaction,guided maintenance process.In the final system verification,the functions of data transmission and virtual model maintenance guidance interaction of the fault diagnosis system are verified.It is proved that the digital twin fault diagnosis system can mine equipment data information,continuously analyze and guide the fault information and maintenance scheme,and provide a new solution in the field of fault diagnosis.