摘要
测试性预计是以UUT测试性指标作为主要研究对象的一项工作,其主要数学工具是概率论。而贝叶斯网络是基于概率论和图论的不确定知识表示模型,在不确定知识和模型表示与推理中表现出卓越的性能。所以可以将贝叶斯网络和测试性预计工作很容易地结合起来。基于贝叶斯网络的测试性预计方法,不但可信度高、建模方便,而且可以很方便地集成到测试、诊断的信息框架内,与其他智能诊断模型进行信息交互。该方法体现了"并行设计"思想,适用于UUT的全寿命周期维护。
Testability prediction is a project taking testability metrics of UUT as subject investigated. The primary mathematic tool of testability prediction is probability theory. And Bayesian networks are model representing uncertainty knowledge based probability theory and graph theory, which has exhibited distinguished performance in representation and reasoning of uncertainty knowledge and graph model domain. So we combine Bayesian networks and testability prediction project together. The testability prediction method base on Bayesian networks is not only modeled conveniently. and is able to bc integrated into information framework of testability and diagnosis. It makes that predictive result from Bayesian method is more believable. The information framework makes testability prediction project easy to exchange information with other projects such as testability allocation project, testability assessment project, reliability engineering and so on. The method embodies the thought of concurrent design. It has been proved to suit to life-cycle health management of UUT.
出处
《弹箭与制导学报》
CSCD
北大核心
2007年第4期232-235,239,共5页
Journal of Projectiles,Rockets,Missiles and Guidance
关键词
测试性预计
贝叶斯网络
信息框架
并行设计
testability prediction
Bayesian networks
information framework
concurrent design
