Research On Fault Diagnosis System Of Rolling Bearing

Rolling bearing is one of the important parts of mechanical equipment.It is widely used in various mechanical devices and equipment.Whether the rolling bearing runs normally or not has a major impact on the operation of the entire machinery.Therefore,it is of great practical significance to develop an analysis system specifically for bearing fault diagnosis.With the development of deep learning technology,this paper has carried out research on bearing fault diagnosis based on the loop gate unit.The main contents of the research are as follows:(1)Analyzed the main structure and working principle of the rolling bearing;determined that the vibration signal of the bearing was used as the input of the fault diagnosis system,and the working state of the bearing(normal,outer ring failure,inner ring failure,rolling element failure)as the output;Analyzed the advantages and disadvantages of the existing machine learning algorithms and deep learning algorithms,and finally decided to design the fault diagnosis model based on the loop gate unit GRU;proposed to use the differential evolution algorithm to optimize the model hyperparameters,and finally completed the overall framework of this article the design of.(2)Aiming at the problem that the standard differential evolution algorithm has slow convergence speed,easy to fall into local optimum and unable to optimize discrete variables,an improved differential evolution algorithm is proposed,and the superiority of the proposed improved differential evolution algorithm is demonstrated through simulation experiments.;By analyzing the performance measurement methods of the multi-class neural network model,the fitness function is constructed,and the improved differential evolution algorithm is used for hyperparameter optimization;simulation experiments verify that the hyperparameters optimized by the differential evolution algorithm proposed in this article are significantly improved The performance of the model was improved,and the expected effect was achieved.(3)Set up a rolling bearing test test bench to collect data on rolling bearing fault vibration data,and effectively expand the data set by adding noise,use the obtained optimal hyperparameters to set up and train the model,and the results show the verification of the model The accuracy has reached 98.31%,which meets actual needs.The paper has 45 pictures,16 tables,and 84 references.