Research On Life Prediction For Rolling Bearing Based On Accelerated Life Testing

As the accessory and basic parts of the machinery industry, rolling bearings arewidely used and strictly demanded, which are considered as the joints of machinerysystem. As the key part of equipments, the failure in rolling bearings will weaken thebattle effectiveness gravely and even cause disastrous fault. The lifetime of rollingbearings is an important statistical index of their durability. Therefore, predicting thelifetime of rolling bearings effectively has great significance to understand the healthcondition of mechanical products and to establish optimal maintenance strategy. Thisthesis centres on the problem of the lifetime prediction of rolling bearings andestablishes a prediction method based on the accelerated life test, and carries outengineering application. The principal researches are as followings：(1) This thesis makes a profound analysis on the fatigue failure mecha nism ofrolling bearings on the base of summarizing the working principle and failure mode ofrolling bearings briefly. Different from the traditional hypothesis that the failure life ofrolling bearings obeys Weibull distribution, this thesis explains why the failure life ofrolling bearings obeys three-parameter Weibull distribution from the analysis of failureand lays a foundation for further data analysis method.(2) This thesis briefly introduces current major parameter estimation methods ofthree parameters Weibull distribution and puts forward a new kind of estimation methodwhich is more suitable for engineering application aiming at the lack of traditionalmethods, and lays a foundation for further statistical method of accelerated life testbased on the competing failure condition.(3) Aiming at the gap of the application of three-parameter Weibull distribution inthe competing failure condition, this thesis establishes three-parameter Weibulldistribution and a statistical method of accelerated life test based on the competingfailure condition. In the process of parameter estimation, this method solves thedisadvantage that the direct maximum likelihood estimation is sensitive to initial valueby introducing a maximum likelihood estimation method based on EM algorithm.Thesimulation results show that the fine estimator can achieve nice performance even in thecondition that the uncertainty image scale factor of failure mode is0.7.(4) On the base of relevant operative standards of accelerated life tests，this thesisestablishes detailed solution of the accelerated life test of rolling bearings and putsforward a method to calculate the upper limit of external load by the formula used tocompute the maximum contact stress on rolling bearings, and proposes some detailedregulations of consistence test of experimental conditions, in order to make sure that thefailure mechanisms of rolling bearings are similar so that the test data from differentsamples is comparable, which can provide theoretical guidance to the implementation of accelerated life test.(5) This thesis use the deep groove ball bearing6205as a subject to implementengineering application and experimental verification. According to the experimentalscheme of the accelerated life test of rolling bearings, it makes statistic analysis of theaccelerated life test of the deep groove ball bearing6205and obtains reliabilityevaluation and life prediction of the rolling bearing6205. The results show that, the testwith competing causes of failure is tally with the actual situation and makes the resultsmore accurate, which verifies the feasibility and effectiveness of the method used in thisthesis.In summary, this thesis studies a method to predict the lifetime of rolling bearingsbased on the accelerated life test with the combination of theoretical analysis andapplication verification. The research achievements will play a promoting role for thereliability evaluation and life prediction of rolling bearings. And it will also providereference significance for the use of the technique of accelerated life test in the area ofreliability evaluation and life prediction of other mechanical products.