Research On Performance Degradation And Life Prediction Method Of Rolling Stock Running Part Bearing

The rapid development of railway transportation has increasingly higher requirements on the stability,comfort and service safety of rolling stock.Rolling bearings are important supporting parts in the running parts of locomotives.Their service conditions,dynamic performance and service life have a crucial impact on the safe,stable and efficient operation of locomotives.Therefore,the research on bearing performance degradation and life prediction method for rolling stock of locomotives has important application value.Based on the identification and tracking of the degradation process of bearing performance and the strong demand for the prediction of the remaining life of the bearing,a statistical analysis of the failure situation of the bearings of high-speed EMUs in China in the past five years is carried out,and the fault diagnosis and remaining life of the locomotive and vehicle bearings The current research status of prediction puts forward reasonable assumptions for the fault diagnosis and remaining life prediction of locomotive and vehicle bearings in China,in order to achieve better bearing monitoring and remaining life prediction,and to ensure the safe operation of locomotives.The main research contents of this article are as follows:(1)Research on the degradation mechanism of rolling bearing performance under the coupling of damage and vibration.A performance degradation model under the coupling of rolling bearing damage and vibration is established,and a numerical solution for the coupled model is proposed.Then,taking the inner surface crack of the inner ring of 6209 deep groove ball bearing as an example,the inner surface crack of the bearing inner ring was used as an entry point to analyze the effect of interference assembly,inner ring crack and rotation speed on the original design load distribution of the bearing,and the load distribution was further analyzed.The effect of the change on the crack propagation,which clearly explains the damage-vibration coupling relationship.Finally,the proposed coupling model and its numerical solution are verified by numerical simulation and experimental verification.The experimental results show that the damage-vibration coupling effect will accelerate the expansion of the crack and cause the degradation process of the bearing performance to be more rapid than the degradation process under the traditional physical model.(2)Research on Tracking and Recognizing Method of Bearing Performance Degradation Based on Symbolic Dynamic Entropy.Aiming at the difference in time scale between the fast-changing vibration component and the slowly-changing damage component in the damage-vibration coupled model,a bearing damage degree tracking and recognition method based on multi-scale symbolic dynamic entropy is proposed.This method uses entropy value change rate,Methods such as mutual information and pseudo-nearest neighbor confirm the key parameters that affect the multi-scale symbol dynamic entropy algorithm,such as the optimal number of symbols,optimal embedding dimension and optimal delay time.Subsequently,a method for extracting sensitive features based on distance assessment technology is proposed.This method calculates the average distance between classes and filters out the sensitive feature components for the recognition of the degree of damage through a defined sensitivity factor,and then uses the support vector machine method Identify the fault location.Finally,through experimental verification,the superiority of the proposed algorithm in tracking and identifying the degraded state is verified.The experimental results show that,compared with the traditional mean value method,mean square error method,kurtosis value method,sample entropy method and permutation entropy method,the proposed multi-scale symbolic dynamic entropy can well identify different defect sizes and by combining Support vector machine can accurately identify the fault location.(3)Research on identification and tracking method of bearing performance degradation based on variational mode decomposition.In view of the fact that the rapidly variable vibration component in the damage-vibration coupled model is very weak in the early stage of defect generation and the background noise and interference in railway conditions are extremely large,this paper proposes a bearing performance degradation and tracking method based on variational mode decomposition technology.Based on the traditional variational modal decomposition,this method uses the energy difference method and the synthetic spectral kurtosis method to realize the optimal selection of the number of modes and the penalty coefficient,thereby improving the accuracy of variational modal decomposition.Subsequently,the comprehensive evaluation criteria for the optimal IMF component based on prior correlation coefficient,singular value,and spectral kurtosis were proposed.The IMF component used to reconstruct the signal was selected through the comprehensive evaluation index,and finally the identification of early failure and the degradation state were realized track.Finally,the proposed method is verified by means of numerical simulation and experimental verification.The experimental results show that the proposed method for identification and tracking of bearing performance degradation based on variational mode decomposition can effectively identify the early bearing under strong background noise The signal synthesized by the fault characteristic frequency at the same time maximizes to contain the fault information in the original signal.(4)Research on Rolling Bearing Remaining Life Prediction Method Based on Physical-Data Driven.Based on the damage-vibration coupled model,a rolling bearing remaining life prediction method is proposed.This method establishes a rolling bearing performance degradation model.The multi-scale symbol dynamic entropy method and the variational mode decomposition method are used to extract the feature vectors Identify different stages of rolling bearing performance degradation by U statistical method and Chebyshev principle.Then the bearing performance degradation index contained in the feature vector is used as input to update the parameters using the data driving method to predict the remaining life of the rolling bearing.The final experimental results show that the physical-data-driven rolling bearing remaining life prediction method has both advantages.It can use the study of the damage mechanism of the physical model and the historical data sub-utilization and data-driven methods to the vibration data collected through real-time Update to achieve a more accurate prediction of remaining bearing life.