基于Jeffreys先验的PSA通用数据贝叶斯处理方法  

Bayesian Method of PSA Generic Data Processing Based on Jeffreys Prior

作　　者：沈志远[1] 陈伟[1] 袁建新[1] 唐秀欢[1] 杨剑[1] 

机构地区：[1]西北核技术研究所强脉冲辐射模拟与效应国家重点实验室,西安710024

出　　处：《核动力工程》2014年第6期84-87,共4页Nuclear Power Engineering

摘　　要：利用Jeffreys先验理论推导出gamma分布和beta分布的Jeffreys先验,在此基础上通过贝叶斯方法得到通用数据分布超参数计算表达式,最后以文献数据为例对失效率分布的超参数进行计算。与经典统计学方法所得结果相比,基于Jeffreys先验的数据处理方法简单易行,可最大限度地保留样本信息,给出的概率分布不确定性较小。Jeffreys priors for gamma and beta distribution was derived after adopting Jeffreys prior theory. Furthermore, the expressions were given to estimate the hyper parameters of generic data distribution using Bayesian approach. Compared to the results using classical statistics theory, the data processing method based on Jeffreys prior appeared much simpler and more concentrated, which reserved most sample information.

关 键 词：贝叶斯方法 Jeffreys先验 概率安全评价 可靠性数据 

分 类 号：TL364[核科学技术—核技术及应用]