基于狄氏先验分布的固体火箭发动机可靠性增长Bayes分析  

Reliability-growth Bayesian analysis for solid rocket motor based on Dirichlet prior distribution

作　　者：刘飞[1] 窦毅芳[1] 张为华 

机构地区：[1]国防科技大学航天与材料工程学院,长沙410073

出　　处：《固体火箭技术》2006年第4期239-242,共4页Journal of Solid Rocket Technology

基　　金：国家"863"项目(2003AA765030)

摘　　要：针对固体火箭发动机研制存在历史信息和专家信息的特点,提出了一种可靠性增长分析的Bayes模型。以狄氏分布作为先验分布,给出了先验分布参数的确定方法。综合利用了先验信息和每一阶段现场试验信息,推导了各阶段可靠性的后验边际分布。给出了当前阶段和后续阶段产品可靠性的Bayes点估计和置信下限,分析了各阶段试验结果对最终产品可靠性估计的影响。最后给出了一个实例,验证了该模型的有效性。Aiming at some history and export information during the development of solid rocket motor （ SRM）, a Bayesian reliability growth model was presented. Taking Dirichlet distribution as prior distribution,the method for determining prior distribution parameters was given, The posterior marginal distributions on reliability were deduced by comprehensively making use of prior information and field test information at every stage. Bayes point assessment and confidence lower limit on product reliability at current stage and later stage were presented ,and the effect of every stage test results on reliability assessment of final product was analyzed. Finally, an example of reliability-growth for SRM was used to illustrate the effectiveness of Bayesian model.

关 键 词：固体火箭发动机 可靠性增长 狄氏分布 贝叶斯分析 后验分布 

分 类 号：V435[航空宇航科学与技术—航空宇航推进理论与工程]