利用贝叶斯网络方法判定整机产品薄弱环节  

Weak Link Analysis of Assembly Product Based on Bayesian Networks

作　　者：王亚辉[1] 曾曼成[1] 

机构地区：[1]中国直升机设计研究所,江西景德镇333001

出　　处：《直升机技术》2015年第1期19-24,33,共7页Helicopter Technique

摘　　要：针对基于外场使用数据的统计分析方法和可靠性分析两类常见的整机产品薄弱环节确定方法进行研究,发现两类方法的应用均受到了限制,提出了一种把传统故障树转化为贝叶斯网络来确定整机产品薄弱环节的方法。该方法用Bayes模型确定贝叶斯网络中各节点发生的条件概率,首先利用Beta分布描述节点发生条件概率的先验分布,参考专家经验,通过极大熵方法给出先验分布中的超参数;然后,再结合整机产品相关试验、类似型号等数据,对节点发生的条件概率做出Bayes估计;最后利用贝叶斯网络的反向推理确定整机产品的薄弱环节。通过光纤陀螺对本方法进行验证,分析结果与工程实际相符合,为整机产品的薄弱环节分析提供了新的思路。A study was made on the weak link ascertain method of assembly products,two common methods based on statistical analysis methods of outfield storage data,based on reliability analysis,it was found that the application of two methods was all restricted,a method that transforms traditional fault trees analysis into Bayesian Networks to find out the weak links of the storage of assembly products was proposed. Which can identifying the Conditional Probability of each node in the Bayesian networks based on the Bayesian model,first,the paper described the prior distribution of conditional Probability of the nodes with beta distribution,reference the expertise,the Hyper parameters of prior distribution was given by the maximum entropy,then,along with the small quantity of storage testing data,Similar models data and so on,the Bayesian estimation of conditional Probability of the nodes was made. And finally the storage weak link was given by backward inference of Bayesian networks.The method was verified by fiber optical gyroscope,and the analysis result corresponds with the actual project. It provided a new route for the weak link analysis of assembly product.

关 键 词：薄弱环节 确定性逻辑门 贝叶斯网络 Bayes模型 BETA分布 极大熵 

分 类 号：TB114.3[理学—概率论与数理统计]