融合多源信息的维纳过程性能退化产品的可靠性评估  

Reliability Assessment of Products with Wiener Process Degradation by Fusing Multiple Information

作　　者：王小林[1] 郭波[1] 程志君[1] 

机构地区：[1]国防科技大学信息系统与管理学院,湖南长沙410073

出　　处：《电子学报》2012年第5期977-982,共6页Acta Electronica Sinica

基　　金：国家青年自然科学基金(No.60904002)

摘　　要：针对维纳过程性能退化产品,提出了一种有效融合先验退化信息、寿命数据以及现场退化数据的可靠性评估方法.首先利用Expectation-Maximization(EM)算法基于先验退化信息和寿命数据信息确定参数的先验分布;其次利用贝叶斯方法对参数进行更新,并在此基础上进行可靠性评估.该方法能根据现场退化数据不断地对可靠性进行更新,实现对产品可靠性的实时评估.最后通过金属化膜电容器可靠性评估实例验证了该方法的适用性和有效性.With respect to Wiener process-degradation products, a method which fused prior degradation information,life data information and current degradation data was proposed to assess their reliability. Firstly, on the basis of the prior degradation infor- mation and the life data information, parameters＇ prior dislributions were established by EM method; secondly, by using Bayesian method, prior distributions could be updated to posterior distributions; then the reliability was evaluated based on the posterior distri- butions. The proposed method can update the reliability evaluation results whenever new degradation data is available, so it is suit- able for real-time reliability assessment.Finally,the validity and practical value of this method is verified by an example,relating to reliability assessment of the metallized film capacitor.

关 键 词：实时可靠性评估 维纳过程 期望最大化算法 贝叶斯方法 金属化膜电容器 

分 类 号：TB114.3[理学—概率论与数理统计]