多源验前信息下结合专家经验的ML-II融合方法  

Fusion of Information of Multiple Sources Based on Expert knowledge and ML-II Theory

作　　者：沈政[1] 师义民[1] 姜祥周[1] 

机构地区：[1]西北工业大学应用数学系,西安710072

出　　处：《科技通报》2008年第4期454-457,共4页Bulletin of Science and Technology

基　　金：国家自然科学基金资助项目(No.70471057);西北工业大学研究生创业种子基金(Z200569)

摘　　要：在Bayes小子样理论中,验前分布的选取是一个关键。针对工程实践中遇到的验前信息的多源性,给出了一种结合专家经验改进的第二类似然原理的多源异总体验前分布的融合方法,以航空电子产品为例,针对产品的失效率,利用Monte-Carlo仿真模拟表明了该方法的有效性。How to acquire proper prior distribution is a key problem in small sample Bayes theory. The situation is that the prior information comes from different sources. A new method based on the second maximum likehood theory and expert knowledge to realize the fusion of information of multiple sources was proposed. Using aerospace electronic product, considering it＇s failure rate, the Monte-Carlo simulation result shows the method is effective.

关 键 词：Bayes定理 ML—II原理 专家经验 信息融合 小样本 

分 类 号：O212.8[理学—概率论与数理统计]