摘要
在诊断空间中如何选取理想诊断是诊断系统面临的一个重要问题 .在实际的诊断过程中 ,人们会利用限制条件排除不太可能的诊断 ,或者利用强制条件选取较优的诊断 .按照这个思想 ,作者提出含约束的基于模型的诊断系统 ,通过增加依赖于应用领域的约束控制诊断空间 ,这是一种能够融入计算过程的选择诊断的机制 ;同时作者在系统拓扑结构的基础上给出了选取理想约束的理论依据 .
The task of fault diagnosis is to, from the observations and the knowledge, decide whether there is a fault or not and also identify the fault. Model-based diagnosis is to perform diagnosis by means of models. The diagnosis process is composed of three iterative stages: hypotheses generation, hypotheses testing and hypotheses discrimination. These stages show that how to select ideal diagnosis from diagnosis space is an important problem with which a diagnosis system must face. As we know, in the real process of diagnosis, one would use a restriction to exclude some less plausible ones, or a coercion to find preferable ones from all possible diagnosis. Based on such an idea, a model-based diagnosis system augmented with domain-dependent constraints is presented. These constraints, which can melt with the process of computing diagnosis, are selection criteria to control the process of diagnosis generation in the sense of filtering some plausible diagnosis or preferring some credible ones. The advantage of our approach is that ideal constraints can reduce half of the candidate diagnosis space. Also this paper addresses how to discover various valuable constraints. By a theorem presenting a sufficient condition when discriminating constraints exist, it is demonstrate that our results build the theoretical foundation for selecting ideal constraints on the basis of system structure.
出处
《计算机学报》
EI
CSCD
北大核心
2001年第2期127-135,共9页
Chinese Journal of Computers
基金
国家自然科学基金! (6 98730 47)
广东省自然科学基金! (980 2 6 0 )资助
