Research On Bearing Performance Degradation Assessment And Prediction

With the continuous development of society,the continuous progress of industry,the rapid growth of urban population,the expansion of city scale,China's urban rail transit has entered a stage of rapid development.Through the research of domestic subway,it found that the main electromechanical equipment is maintained by regular maintenance or after-failure repair.The traditional FAF concept can not adapt to the new development and new requirements of the environment.The study of PAP degradation evaluation and prediction based on PAP is of great theoretical and practical value.Aiming at the insufficiency of maintenance of electromechanical equipment,this paper presents a complete new theoretical method of bearing performance degradation assessment and prediction based on the key components of electromechanical equipment-bearing.The validity of the method is validated by bearing life cycle data.The key to performance degradation assessment and prediction lies in the new method of feature extraction,sensitive feature index and new and effective bearing performance degradation evaluation and prediction model.For the feature extraction,in this paper,the local mean decomposition algorithm proposed in the field of signal analysis and processing has been studied,and the problem of phase difference and sliding step selection in the algorithm step has been studied.An improved LMD algorithm with smaller decomposition error and higher computational efficiency is proposed.Then,the improved LMD algorithm is combined with information entropy theory,and LMD energy spectrum entropy is put forward as an indicator of bearing performance monitoring.In view of the existing characteristic indexes,the problem of the capacity of quantitative degradation of equipment degeneration is insufficient.Because the intermediate process from normal to fault is extremely unstable,the LMD energy spectrum entropy is a new index which reflects the degradation of bearing performance.It can not establish the corresponding relationship with the traditional indexes such as the RMS,peak value and so on.It will change because of the difference of operating conditions.So the theoretical method of calculating the critical value of degradation is put forward.Then the paper puts forward the regression evaluation model based on logical regression and expands the existing research ideas and research methods.The prediction model based on the extreme learning machine is put forward,which is a new application in this field.Through the run-to-failure experimental data of bearing,both of which are well evaluated and predicted.Using the theory of this paper,the paper designs the bearing maintenance system software for the key equipment of urban rail transit,which has high application value and practical significance.Finally,this paper designs the bearing maintenance system software for the key equipment of urban rail transit,which has high application value and practical significance.