双应力恒加试验Weibull分布型产品置信可靠性评估模型  

Confidence reliability evaluation model of dual constant-stress accelerated life test for Weibull distribution product

作　　者：梁红琴[1] 冯雪峰 唐家银[2] 刘奇锋 LIANG Hongqin;FENG Xuefeng;TANG Jiayin;LIU Qifeng(School of Mechanical Engineering, Southwest Jiaotong University, Chengdu 610031, China;School of Mathematics, Southwest Jiaotong University, Chengdu 611756, China)

机构地区：[1]西南交通大学机械工程学院,四川成都610031 [2]西南交通大学数学学院,四川成都611756

出　　处：《系统工程与电子技术》2020年第9期2140-2148,共9页Systems Engineering and Electronics

基　　金：中国铁路总公司科技研究开发计划项目(2017J003-H);中央高校基本科研业务费(2682017ZDPY13);教育部人文社会科学研究规化基金(20XJAZH009)资助课题。

摘　　要：为了有效快速评估高可靠性长寿命产品在正常应力条件下的可靠性,提出了一种双应力恒加试验Weibull分布型产品置信可靠性评估模型。首先,采用加速寿命试验(accelerated life test,ALT)技术,建立广义Eyring-Weibull可靠性模型,假定试验各应力水平组合下Weibull分布的形状参数相同,且尺度参数与各应力水平组合间呈对数线性关系。其次,给出了Weibull分布定时截尾双应力恒加试验的极大似然估计(maximum likelihood estimation,MLE)方法、与似然函数相关的Fisher信息矩阵以及模型参数的渐近协方差矩阵,构造了模型参数和一些可靠性指标的渐近置信区间。最后,通过仿真算例证明了所提方法的可行性。In order to evaluate the reliability of high reliability and long life products under normal stress conditions effectively and quickly,a confidence reliability evaluation model of dual constant-stress accelerated life test(DCSALT)for Weibull distribution product is proposed.Firstly,the generalized Eyring-Weibull reliability model is established by using the accelerated life test(ALT)technology.In this model,it is assumed that the Weibull distribution has constant shape parameter and a log-linear relation between its scale parameter and each stress level combination.Secondly,a maximum likelihood estimation(MLE)method for DCSALT with fixed time censoring of the Weibull distribution,the Fisher information matrix realted to the likelihood function,and the asymptotic covariance matrix of the model parameters is presented.Furthermore,the asymptotic confidence intervals of the model parameters and some reliability indexs are constructed.Finally,the simulation example demonstrates the feasibility of the proposed method.

关 键 词：双应力恒加试验 加速方程 可靠性 WEIBULL分布 Fisher信息矩阵 

分 类 号：TB114.3[理学—概率论与数理统计]