基于混合诊断贝叶斯网络的测试性不确定性建模与预计  

Hybrid Diagnostic Bayesian Networks Based Uncertain Testability Modeling and Prediction

作　　者：王宝龙[1,2] 黄考利[2] 张亮[1] 郝莉娜[1] 

机构地区：[1]北京航天飞行控制中心,北京100094 [2]解放军军械工程学院,石家庄050003

出　　处：《弹箭与制导学报》2013年第2期177-180,F0003,共5页Journal of Projectiles,Rockets,Missiles and Guidance

摘　　要：针对目前测试性预计方法缺乏对不确定性全面考虑的现象,从测试和故障两个方面分析了测试性预计过程中的不确定性问题,基于贝叶斯网络测试性模型对不确定性测试问题进行了建模与分析,基于混合诊断模型对故障不确定性问题进行了建模与分析,并最终将贝叶斯网络测试性模型与混合诊断模型相融合,给出了一种兼容IEEE Std 1522的基于混合诊断贝叶斯网络模型的测试性指标预计方法,测试性指标预计结果的可信度明显提高。In view of ignorance of uncertain problem existed in test and diagnosis, current testability prediction method tends to be inaccu-rate. The hybrid diagnostic Bayesian networks can model the uncertain relationship between tests and failure sources, failure modes and functions. In the paper, a method of hybrid diagnostic Bayesian networks was given based on uncertain testability modeling and prediction, which is compatible with IEEE Std 1522. The model of hybrid diagnostic Bayesian Networks is a kind of enhanced diagnostic inference model which can provide testability prediction solution with high accuracy.

关 键 词：贝叶斯网络 混合诊断模型 测试性 不确定性 预计 

分 类 号：TP202[自动化与计算机技术—检测技术与自动化装置]