基于多源数据融合Bayes修正的可靠性综合评估模型  

Reliability Comprehensive Evaluation Model Based on Multi-source Data Fusion Bayes Correction Model

作　　者：曹玲 唐家银[1] 曹宇宇 何平[1] Cao Ling;Tang Jiayin;Cao Yuyu;He Ping(School of Mathematics,Southwest Jiaotong University,Chengdu 611756,China)

机构地区：[1]西南交通大学数学学院,四川成都611756

出　　处：《甘肃科学学报》2021年第3期35-41,60,共8页Journal of Gansu Sciences

基　　金：中国铁路总公司科技研究开发计划项目(2017J003-H);四川省统计科学研究计划项目(2016sc50);中央高校基本科研业务费专项项目(2682017ZDPY13)。

摘　　要：基于Bayes统计理论,融合同类产品的加速退化试验信息和待估产品工作环境中的正常退化信息,进行信息融合可靠性综合评估。针对性能退化过程服从Gamma过程的产品,考虑个体退化之间的差异,将Gamma过程参数进行随机化处理;考虑正常工作环境与加速试验工况的外载环境差异,建立带有修正系数的广义寿命-应力响应Bayes可靠性模型;根据正常退化和加速退化数据,利用加速模型,折算出各未知参数先验分布的数据样本,采用A-D检验法及极大似然估计确定参数集的先验分布;采用MCMC方法求解各分布参数的后验均值以及后验概率密度曲线。通过算例验证了该类模型的可行性和有效性。Based on Bayes statistical theory,the accelerated degradation tests of similar products and the normal degradation in the working conditions of the product to be evaluated were integrated in this paper.And the reliability estimation of the product was carried out to comprehensively evaluate the information fusion.The products whose performance degradation process obeys generally the Gamma process in which the difference between individual degradations was considered,and the Gamma process parameters are randomized.By using the difference of the stress external environment between the normal working environment and the laboratory accelerated tests conditions,a generalized life-stress response model with a correction factor was established.Subsequently,the acceleration model was used to calculate the data samples of the prior distribution of each unknown parameter,and the A-D test and the maximum likelihood estimation were used to determine the prior distribution of the parameter set.The posterior mean value and posterior probability density curve of each distribution parameter are solved by MCMC method.The test verified the feasibility and effectiveness of this model.

关 键 词：Gamma过程 Bayes理论 加速退化 修正系数 MCMC方法 

分 类 号：TB114.3[理学—概率论与数理统计] O213.2[理学—数学]