摘要
健康评估是综合运载器健康管理的关键技术之一。针对健康评估的需求,提出适合于健康评估功能要求的五要素建模理论,研究了基于面向对象的贝叶斯网络(OOBN,Object-Oriented Bayesian Networks)的健康评估实现技术,提出了五要素结构化OOBN健康评估推理模型的建模方法,鉴于贝叶斯网络(BN,Bayesian Networks)推理计算属于NP-hard难题,难于工程化应用,提出了适用于结构化BN模型的,以变化证据信息为导向的,基于贝叶斯球和交叉熵测度的变模型快速推理算法,Monte-Carlo仿真实验验证了算法的有效性,最后以变速箱健康评估为例,说明本文所提健康评估理论的有效性。
Health assessment is one of the key technologies of integrated vehicle health management （ IVHM）. To solve health assessment requirement, a new five elements modeling theory was proposed. The health assessment implemen- tation based on Object-Oriented Bayesian Networks （OOBN） was studied, and the modeling approach on five elements structured OOBN health assessment models was also proposed. In view of Bayesian Networks （BN） inference is NP-hard and difficult to engineering applications, a quick inference algorithm based on Bayes-Ball and cross entropy was proposed, which is change-evidence-oriented and applicable to the structured BN models. The Monte-Carlo simulation result shows the quick inference algorithm is efficient. The gearbox health assessment experiment shows the practicability of the proposed health assessment theory.
出处
《宇航学报》
EI
CAS
CSCD
北大核心
2009年第4期1711-1721,共11页
Journal of Astronautics
基金
总装"十一五"预研项目
关键词
综合运载器健康管理
健康评估
建模
面向对象的贝叶斯网络
交叉熵
贝叶斯球
Integrated vehicle health management
Health assessment
Modeling
Object-Oriented Bayesian networks
Cross entropy
Bayes-ball
