Prediction Of Remaining Useful Life And Maintenance Decision For Reliable Operation Of Mechanical Rolling Bearings

As the most widely used and most easily damaged parts in rotating machinery,the key to ensure the healthy and efficient operation of machinery equipment is to accurately recognize its operation status,predict its healthy service time,and then provide effective maintenance decisions.Considering the ability of the proportional hazards model to effectively combine the uniformity of equipment fault rules with the particularity of individual state deterioration process,the ability to timely update equipment reliability based on real-time state monitoring data,and the ability to provide a basis for bearing life prediction and failure prevention strategies.The following contents are studied in this paper:(1)This paper briefly introduces two forms of weibull distribution,introduces them into proportional hazards model to form weibull proportional hazards model(WPHM),and analyzes the specific forms of WPHM,revealing the reason why the expression of WPHM is a form of two parameters.Discuss the discrete value rule of WPHM covariable and the parameter solution method:(1)Put forward the discrete covariable "step hypothesis",analyze the drawbacks of the "right step" rule,and confirm that the value of the covariable is "left step".(2)The step by step calculation principle of WPHM parameters is explored.In the case that the parameters of the benchmark risk rate function are known,the maximum likelihood function is taken as the objective function according to the optimization theory,and the maximum likelihood estimation model of WPHM is established.The improved shrinkage factor particle swarm optimization(MCFPSO)algorithm is used to realize the fast convergence of the remaining parameters.(2)The key to judging the running life of bearing is to predict the development trend of covariates in WPHM.Based on the grey theory,the PGM(1,1)prediction method is constructed by optimizing the internal parameters of the original grey model GM(1,1)with particle swarm optimization(PSO).Then the fuzzy mathematics theory is introduced and the historical relative error generated by the original data fitting is fuzzily processed.Finally,a new combined prediction model PGFM(1,1)is formed by combining markov chain prediction theory to make up for the poor prediction effect of grey model on state jump,so as to accurately predict the development trend of bearing covariates.(3)Using the rolling bearing life cycle degradation test provided by the intelligent maintenance center of the university of Cincinnati,the relationship between the actual residual life,WPHM predicted residual life and PGM + WPHM predicted residual life was compared,and the validity of the proposed life prediction method was verified.Then,based on PGM+WPHM,the maintenance decision of bearing maximum availability is applied,and the three-stage maintenance timing prediction decision diagram is obtained,which provides reasonable suggestions for bearing maintenance time point and effective allocation of condition monitoring resources.(4)The main failure mode of bearing is the fatigue failure of the structure under the action of periodic variable amplitude load.Therefore,at the end of the paper,the fatigue life calculation process is introduced with the route of alternating stress cycle extraction--average stress correction--cumulative damage criterion.It provides a method for predicting the fatigue life of rolling bearing.