数学物理学报 ›› 2003, Vol. 23 ›› Issue (4): 474-484.

• 论文 • 上一篇    下一篇

用于指数可靠性增长模型的一类新的先验分布

 吴启光, 赵勇辉   

  1. 中国科学院系统科学研究所  北京 100080 中国科学院应用数学研究所  北京 100080
  • 出版日期:2003-08-25 发布日期:2003-08-25
  • 基金资助:

    国家自然科学基金(19871088, 19631040)资助

A New Family of a prior Distributions for Exponential Reliability Growth Models

 TUN Qi-Guang, DIAO Yong-Hui   

  • Online:2003-08-25 Published:2003-08-25
  • Supported by:

    国家自然科学基金(19871088, 19631040)资助

摘要:

该文提出了可用于指数分布产品四种可靠性增长试验方案的一类新的先验分布. 这类先验分布以条件分布形式给出, 它适合可靠性增长试验中的各种情况. 各阶段的条件均值和条件方差的表达式被获得, 先验分布的形式与它们的参数间的关系被讨论. 这些结果有助于与专家意见相结合.本文还给出试验末尾产品可靠性的后验密度, Bayesian估计和Bayesian下限.

关键词: 可靠性增长试验, 指数分布, Bayesian推断

Abstract:

This paper introduces a new family of a prior distributions for four reliability growth testing scenarios with exponential failure data. The proposed prior is of conditional form, which can accord well with various actual situations in reliability growth tests. The expressions of corresponding conditional means and variances for all stages are obtained, and the relationship between the shape of a prior distributions and their parameters are iscussed. These results are helpful for better incorporation of expert opinons. The posterior density and Bayesian estimators and Bayesian lower bound of the reliability at the end of the test are also given.

Key words: Reliability growth test, Exponential distribution, Bayesian inference