智能世界的建模与诊断

Modeling and Diagnosis of the Intelligent World

统一的抽象建模框架以及形式化表示可以帮助实现自动推理.随着物联网技术的发展,物理世界中嵌入了各种智能对象,改变了物理世界的部分特征,增加了建模和推理的复杂性.根据物联网带来的智能世界的特征,在知识重构与抽象模型(KRA模型)的统一建模框架基础上,提出了可区分的知识重构与抽象模型(dKRA模型).该模型通过3个相互关联的子模型及其之间的关系来表示智能世界,并给出相关定义和定理说明在所提出的模型框架内,可以将基于模型的诊断过程限制在一个(或多个)子模型中.研究内容侧重于系统设计阶段的模型验证,分别从理论和实验角度分析了基于智能世界dKRA模型的诊断过程时间效率的提高(与基于智能世界KRA模型的诊断过程相比).

Unified abstraction modeling framework and the formal representation help to realize automated reasoning. With the rapid development of the Internet of Things （loT）, various smart or intelligent ob}ects are embedded in the physical world which partly changes the characters of the composed entities. Our familiar world is accordingly changing into an intelligent world which greatly increases the modeling and reasoning complexity. In this paper, based on the unified knowledge reformulation and abstraction （KRA） model framework presented by Saitta and Zucker, we propose a distinguishable knowledge reformulation and abstraction （dKRA） model according to the characters of the intelligent world. Different from the KRA model which represents the intelligent world in a unified way, the proposed dKRA model can be formalized through three interrelated sub-models and their relationships. We give some definitions and theorems based on the correlative concepts of model- based diagnosis to show that the diagnosis process can be limited in one （or more than one） of the sub- models. By focusing on the model verification during the phase of system designing, the presented diagnosis algorithm running on the dKRA model of an intelligent world is analyzed from both theoretical and experimental perspectives to show the improvement on the diagnosis process running directly on the KRA model of the same intelligent world.