专家系统中一类新的近似推理模型ARM

A New Appoximate Reasoning Model for Expert Systems

本文提出专家系统中的基于模糊语言逻辑的一类新的近似推理模型ARM。在ARM中,我们把领域专家知识表示的基本单元表征为一个三元组(P,β,LTV),其中P是一个模糊断言;β∈[0,1]是可信度;LTV是自信度,即语言真值。作为知识表示基本单元的三元组(P,β,LTV)将事实、模糊关系(模糊谓词)、模糊产生式规则等结合起来,构成一个模糊推理网络。基于ARM的专家系统中的推理机制包括正向推理,逆向推理和双向推理。给出了在正向推理、逆向推理和双向推理中模糊断言的可信度、自信度和真值的传递运算方法。这类近似推理模型ARM已用于建造“煤矿生产技术管理专家系统”的推理机制。目前,就这个专家系统运行的实例来看,其推理结果与领域专家的观点十分一致。

In this paper a fuzzy linguistic logic based approximate reasonging model(ARM) for expert systems is introduced. In ARM a basic unit of the domain-specific knowledge representation is characterized by a triad (P, β, LTV) in which P is a fuzzy assertion; β∈[0,1] is the degree of confidence; LTV is the degree of self-confidence, that is,the linguistic truth. The triad (P, β, LTV), as a basic unit of knowledge representation, combine facts, fuzzy reasoning networks. The ARM based inference mechanism for expert systems include forward inference, inverse inference and bidirectional inference. The transferable operations' methods of the fuzzy asserts' the degree of confidence, the degree of self-confidence and the truth in forward inference, inverse inference and bidirectional inference are given. The ARM has been applied to build the inference mechanism of an expert system for coalmine's production and technical management. The expert system is on trial with examples at present, and the results from the examples are perfectly consistent with the views of domain experts.