含缺省约束的基于模型诊断方法

Model-based Diagnosis with Default Constraints

目前已有的诊断系统一般都默认有完备而可靠的模型,但在实际中常无法实现.因此,在目前已知的对待诊断设备的可靠描述之外,提出了可加入假设性的信息作为对无法完备模型的补充,并且这些假设采用了从规范缺省理论中的规则转化而成的公式的形式.进一步给出了在加入这些假设后的情况下,含缺省约束的基于模型的诊断、中心诊断、蕴含、蕴含式及本原蕴含等概念,证明了含缺省约束的基于模型的中心诊断与缺省本原蕴含的直接关系,并指出这种刻画即为含缺省约束的基于一致性中心诊断和中心溯因诊断的刻画的一般化情形,从而将理论与实现联系了起来.

A complete and sound model was considered as default in model-based diagnosis systems,but it was often difficult to reach such a criteria in practice.To solve the problem,this paper added some assumptions into the sound descriptions of the device to be diagnosed as a supplement of the incomplete model,and all the assumptions were described by formulae transformed from default rules of normal default theory.The concepts of model-based diagnosis,kernel diagnosis,implicant,implicate and prime implicant of the new model-based diagnosis system containing default constraints were also expounded with the proof of the straight link between model-based diagnosis and the prime implicants in this assumption-added model-based diagnosis system.In the end,this paper indicates that the link is the generalization of kernel consistency-based diagnosis and kernel abductive diagnosis,which connects the theory with implementing.