基于无信息先验分布的高可靠性产品可靠性评估方法研究  

Reliability Evaluation Method on High Reliability Product Based on No Informative Prior Distribution

作　　者：黄景德[1] 刘薇[1] 杨勇 

机构地区：[1]海军大连舰艇学院舰炮系,辽宁大连116018 [2]东海舰队,浙江宁波315121

出　　处：《测试技术学报》2015年第2期100-104,共5页Journal of Test and Measurement Technology

基　　金：国家自然科学基金资助项目(60971092)

摘　　要：高可靠性产品检验项目中存在大量的无失效数据,使得整机的可靠性评估更加困难,为准确评估高价值产品的可靠性,在多层Bayes方法基础上,重点研究了在小子样的情况下,没有验前信息或者验前信息不足以确定先验分布时,充分利用生产和研制过程中零部件及单元的各种信息对系统可靠性进行精确评估,确定了二项分布无信息先验分布的方法,探讨了综合评判法在多层Bayes方法中的应用,为可靠性要求高、结构复杂、单元信息分布多样化复杂系统的可靠性评估奠定了基础.该方法适用于造价昂贵的高价值产品可靠性评估,在工程上有一定的实用和推广价值.Large number of non-failure data exist in high reliability test items makes it more difficult to assess the reliability of machine. In order to assess the reliability of high-value products exactly, the re- search was focused on the small sample case based on the multilayer Bayes method. When there is no pri- or information or prior information is not sufficient to determine the prior distribution, various informa- tion of parts and units of production is made full use to accurately assess the system reliability. The method of non informative prior distribution of the binomial distribution was identified and the applica- tion of the comprehensive evaluation method in the multilayer Bayes method was discussed. It has laid the foundation for the high reliability requirements, complex structure, and unit information distribution in complex system reliability evaluation. The method is suitable for reliability assessment of the high value expensive product and it has certain practical and promotional value in engineering.

关 键 词：无信息先验分布 多层Bayes方法 高可靠性产品 可靠性评估 

分 类 号：TB114.3[理学—概率论与数理统计]