复杂仿真系统可信度评估框架研究  

Study on Credibility Evaluation Framework of Complex Simulation Systems

作　　者：王石[1] 伍丁红[2] 瞿亮[1] 戴金海[3] 

机构地区：[1]湖南大学电气与信息工程学院,湖南长沙410082 [2]湖北经济学院计算机学院,湖北武汉430205 [3]国防科技大学航天与材料工程学院,湖南长沙410073

出　　处：《计算机仿真》2012年第4期116-122,145,共8页Computer Simulation

摘　　要：关于复杂仿真系统VV&A可信度评估问题,模型可信度层次结构的划分和仿真过程不同阶段对应模型可信度之间的依赖关系,底层模型单元的可信度的确定方法各不同。通过采用贝叶斯网的概念与方法,对复杂建模与仿真系统全生命周期过程的可信度评估框架进行了研究。贝叶斯网的结构可以描述开发过程的每一阶段模型的层次结构以及不同开发阶段对应模型的可信度的依赖关系;节点变量的主观概率值是模型的可信度。通过分析和验证,表明贝叶斯网的可信度评估框架与方法,不但合理、正确而且有效。Credibility evaluation of complex simulation system is the key of VVA,and the crucial problem is the hierarchical structure of models＇ credibility,the dependency relationships of the corresponding models between different phases of simulation development process and credibility evaluation of the bottom layer＇s model units.In this paper,the credibility evaluating framework of the whole lifecycle of complex modeling and simulation was studied using Bayes networks＇ concepts and methods.Bayes networks can easily describe the hierarchical structure of models＇ credibility and the dependency relationships of the corresponding models between different phases of simulation development process,and moreover,the probability of each node variables represents the credibility of the corresponding model.Through discussion and analyses,it shows that the evaluation method based on Bayes networks is correct and efficient.

关 键 词：建模与仿真 仿真可信度 主观概率 贝叶斯网 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]