整机产品失效机理危害度定量分析方法

针对整机产品失效机理多而复杂、失效数据少、主失效机理分析为定性方法的现状,提出了一种基于Bayes的失效机理危害度定量分析方法.首先利用Dirichlet分布描述产品的先验信息,结合现场失效数据确定失效频率的Bayes后验分布函数,然后通过Slice抽样方法获得失效频率的满条件分布,进而利用Gibbs抽样得到其Bayes估计,从而实现了各失效机理的定量分析.最后在某型光纤连接器失效机理已分析清楚的前提下,分别用FMMEA(Failure Modes,Mechanisms and Effects Analysis)及Bayes方法确定了其主失效机理,结果表明基于Bayes的失效机理危害度定量分析方法考虑了现场未出现的失效机理危害度,其确定的主失效机理比FMMEA结果更具有可信度.

Quantitative analysis method for mechanism severity of whole product

Due to qualitative method was used to analyze whole products with characters of complex failure mechanism and small samples, Bayesian method was proposed for quantitative analysis of failure mechanism severity. Firstly, Bayesian posterior density function was determined by prior information described by dirichlet distribution and failure data. Then full condition distribution of failure frequency and Bayesian estimation were obtained by Slice samplings and Gibbs samplings separately. Furthermore, failure mechanism severity was analyzed quantitatively. Finally, under the premise of failure mechanisms of an optical fiber connector have been analyzed clearly, the main failure mechanisms were confirmed by failure modes, mechanisms and effects analysis(FMMEA) and Bayesian method respectively. The results showed that severity of the failure mechanism which did not appeared on site was considered by Bayesian quantitative analysis method, the main failure mechanisms were more credible than the results based on FMMEA.