Prediction Of Rolling Bearing Residual Life Based On Feature Fusion

Rolling bearing is the supporting element in mechanical equipment,widely used in all fields of national economy.Failure of the rolling bearing will result in the failure of the entire mechanical system.Fault diagnosis and life prediction of rolling bearing are beneficial to the realization of equipment health management so as to improve the overall efficiency of mechanical equipment.In this thesis,several time-domain characteristics of vibration signals are extracted from experimental data of rolling bearings accelerated degradation.By using D-S evidence theory,multiple time-domain features are fused into one health index to characterize the degradation state of rolling bearings;In view of the characteristics that the rolling bearing does not degrade obviously in the early stage and rapidly in the later stage,a phase division method based on the 3σ principle was proposed,which successfully divided the degradation process of the rolling bearing into healthy stage and decline stage;In the bearing decay stage,Box＿Cox transformation is used to further enhance the linear correlation between the data,and the least square method is used to determine the parameters of the linear degradation model,so as to realize the prediction of the remaining life of the bearing.In the process of Box＿Cox transformation,a method of parameter prediction based on SVR is proposed to improve the quality of Box＿Cox transformation.Compared with the feature fusion method based on PCA,the feature fusion method based on D-S evidence theory has obvious advantages in health indicator noise,rationality of stage division of rolling bearing and accuracy of prediction model.In terms of residual life prediction results,the regression models established by different prediction methods are compared,and the experimental results show that the residual life prediction method based on the improved Box＿Cox transformation proposed in this thesis can achieve the accurate prediction of residual life of rolling bearings.