Reliability Research Of Rolling Bearing Based On Poor Information And Zero-failure Data

Rolling bearing is one of the most important parts in rotating machinery and equipment.With the increasing complexity of modern machinery and equipment,as well as the requirements of high speed and heavy load,the reliability requirements of rolling bearing are getting higher and higher.Rolling bearing is a performance degradation product.In some application scenarios,the cost and difficulty of obtaining its failure data are getting higher and higher.Based on its small amount of working data,reliability evaluation is needed under the condition of poor information and zero-failure data.E-Bayes method and multilayer Bayes method are widely used in evaluating the reliability of rolling bearings based on the poor information and failure-free data.An important issue in the evaluation process is to obtain the prior distribution of parameters from the prior information,in which the determination of super parameter c will have a direct impact on the evaluation results.At present,there are few studies on the correlation between super parameter c and reliability.Based on the comparative analysis of E-Bayes and multi-layer Bayes evaluation methods,this paper proposes an improved grey correlation model to analyze the influence of super parameter c on the evaluation results.On the basis of determining the evaluation method and super parameter c,the reliability estimation value is calculated,and the residual life is predicted by using the non-equidistant grey model.It provides a reference for the selection of reliability evaluation methods and life prediction of rolling bearings under the condition of poor information and invalid data.The main research work and innovative achievements are as follows :(1)Reliability evaluation method of rolling bearing based on variable prior distribution area.The zero-failure data model of rolling bearing is constructed.By changing the size of the prior distribution interval,the reliability of the rolling bearing is estimated.Taking the simulation data of high-speed train rolling bearings as an example,we carried outthe reliability evaluation of rolling bearing,and the reliability estimation value of rolling bearings is obtained.By comparing and analyzing the classical estimation,it is concluded that the difference between the classical estimation and the traditional Bayesian estimation that fully considers the prior distribution is obtained when the prior information is not fully used.(2)Analysis of rolling bearing reliability evolution based on grey correlation degree.Grey correlation analysis was conducted on the reliability estimation values of multi-layer Bayes and E-Bayes reliability evaluation methods for rolling bearings.The change of super parameter c in the two Bayesian estimation methods is compared for quantitative discussion.The evaluation results of different parameters were analyzed by grey absolute correlation degree,grey relative correlation degree and grey comprehensive correlation degree,and the change rule and relationship were determined.By changing the size of the resolution coefficient,the traditional Deng ’ s correlation degree and the improved Deng ’ s correlation degree are discussed to evaluate the reliability.It is concluded that the E-Bayes estimation method is better in the case of no failure data of rolling bearings,and is less affected by the super parameter c.The zero-failure data obtained from the timing censoring test of rolling bearings are verified,indicating that the analysis method is feasible.(3)An unequal interval grey prediction method for residual life of rolling bearings is proposed.Aiming at the problem of unequal interval in the test of rolling bearing timing censoring,an unequal interval grey prediction method is proposed.According to the rolling bearing failure-free data,the E-Bayes reliability estimation is calculated,and the non-equal interval reliability estimation is transformed into the equal interval reliability estimation,and then the grey prediction model GM(1,1)is predicted.Finally,the formula of rolling bearing life prediction is obtained,which can predict the remaining life of rolling bearings at any time.This paper studies the reliability of rolling bearings,mainly based on the poor information to evaluate the reliability of the non-failure state of rolling bearings.The research content of this paper enriches the reliability research methods of rolling bearings,and provides a theoretical basis for the reliability research of bearings and the application of information-poor theory.