摘要
针对装备研制过程中产生的大量试验和调试数据,提出采用Bayesian网挖掘各组成单元间的依赖关系,并对Bayesian网学习中基于信息论的方法进行了改进,使确定网络拓扑结构的过程更加客观.学习得到Bayesian网后,分析了其在失效源判定和发现设计缺陷等方面的应用.
A lot of experiment data are produced during the process of equipment development, therefore exploiting these data by the data mining instrument becomes an important issue. Bayesian network is then proposed. And the approach, which is used in the study course of Bayesian network and is based on the information theory, is improved by the statistics knowledge, and the topology of Bayesian networks is constructed. Combined with the concept of extended Markov blanket, a method to determine the faulty unit is presented. The conditional probability table of Bayesian network can also assist the personnel in finding out the defect of design.
出处
《海军工程大学学报》
CAS
2004年第2期99-102,共4页
Journal of Naval University of Engineering
