Research On The Prediction Of Rolling Bearing Performance Degradation

During the rapid development of modern industry,the safety and reliability of mechanical equipment are particularly important.As an important part of rotating machinery,the health of rolling bearings directly affects the safe operation of the equipment,so the performance degradation prediction of rolling bearings research has great significance.Taking rolling bearings as research objects,this paper proposes a method for predicting the performance degradation of rolling bearings based on the Krill Herd(KH)algorithm parameter optimization Hybrid Kernel Function Support Vector Regression Machine(HKF-SVR)model and Deep Belief Network-Extreme Learning Machine(DBN-ELM)fusion model.First,the method of combining the Kernel Joint Approximate Diagonalization of Eigenmatrices(KJADE)and the two types of models is used to extract the performance degradation index of the rolling bearing.Then,the traditional Support Vector Regression Machine(SVR)parameters are uncertain and the prediction accuracy is not enough,and because the rolling bearing degradation process is a non-linear and complex process,the original single kernel function SVR cannot effectively implement the rolling bearing degradation assessment.a method for predicting the performance degradation of rolling bearings is proposed which was based on HKF-SVR.The parameters of the HKF-SVR model are optimized using the KH,and the performance degradation prediction of rolling bearings was predicted using the optimized model.Experimental results show that this method has a better accuracy than the traditional SVR?BPNN and ELM in the prediction of rolling bearing performance degradation.Secondly,a method based on the DBN-ELM fusion model to predict the performance degradation of rolling bearings is studied.In view of the strong ability of DBN to extract features from the original data and the fast learning speed and good generalization performance of ELM,a method for predicting the performance degradation of rolling bearings combining the advantages of both DBN and ELM is proposed.Using DBN as a feature extractor,train the original performance degradation index of the bearing,extract the sensitive features that can represent the degradation status of the rolling bearing performance,then input the sensitive features into the ELM model for training and predict the rolling bearing performance degradation.Comparing the experimental results with the results of the single DBN and ELM methods,the results show that this method is better than the other two methods.The data of rolling bearing vibration research in this article are from the rolling bearing fatigue test data of Cincinnati University and our laboratory.Two kinds of data are used to verify the effectiveness of the two methods proposed in this paper.Experimental results show that the two methods proposed in this paper are effective and superior in the prediction of rolling bearing performance degradation,and have higher prediction accuracy than other methods.It can be used to predict the degradation of rolling bearings,providing a reference for rolling bearing life prediction and equipment maintenance.