基于混合Beta先验分布的成败型产品的可靠性评估  

Reliability Evaluation Based on Mixed Beta Priori Distribution in Binomial Case

作　　者：赵勇[1] 刘建新[1] 牛青坡[1] 

机构地区：[1]中国空空导弹研究院,河南洛阳471009

出　　处：《航空兵器》2014年第6期59-61,共3页Aero Weaponry

摘　　要：成败型产品可靠性的小样本评估方法,把历史试验数据作为Bayes估计方法的先验信息,有效地减少了样本量。但是传统的Bayes方法过度依赖验前信息,当新产品有改进时造成鉴定结果的冒进。针对过度依赖先验信息问题,本文提出了使用混合Beta先验分布,引入继承因子,通过调整继承因子的大小来控制对验前信息的依赖程度,考虑了成败型产品的变化对评估结果的影响。采用该方法对某导弹的可靠性进行了评估。Evaluation method for small sample of success /failure product reliability takes historical test data as prior information of Bayes, effectively reducing the quantity of samples.But over-reliance on prior information in the traditional Bayes method, will get the aggressive identification results when new products are improved.To deal with the problem of over-reliance on prior information, this paper puts for-ward using Bages evaluation based on mixed Beta prior distribution, which introducing inheritance factor. Thus the dependence on prior information is controlled by adjusting the inheritance factor, and the effect of changes of success/failure product on evaluation result is considered.With this method, the missile re-liability is assessed.

关 键 词：成败型产品 混合Beta先验分布 继承因子 可靠性 

分 类 号：TJ760.1[兵器科学与技术—武器系统与运用工程]