模糊推理的合成规则及其合成运算的选择研究

RESEARCH ON SELECTION OF COMPOSITIONAL INFERENCE RULES AND COMPOSITIONAL OPERATORS OF FUZZY INFERENCE RULES

本文首先提出当知识库中含有多条取自于同论域上的模糊推理规则时，应用通常的六条合成推理规则所推得的结论将随着模糊推理规则数的增多而越来越偏离真实度这一问题。本文针对该问题对通常的六条合成推理规则及其三角模下的六条推广合成推理规则进行了理论比较研究，并对合成运算进行了比较研究。研究表明：合成推理规则（或Ｒ￣４）以合成运算ｍａｘ－Ｔ０是克服上述问题的最佳选择。这一重要结果对于模糊知识库的设计具有指导意义。

In this paper, a problem is presented that with more fuzzy inference rules based on the same universe of discourse and contained in fuzzy knowledge base, the conclusions obtained by six compositional inference rules respectively will be more derivative from truth. Based on triangular norms, six compositional inference rules are generalized. To solve the above problem efficiently, six compositional inference rules and compositional operators are compared and studied.An important conclusion is obtained'that the selection of the compositional inference rule R'(orR') and the compositional operator max--T,' is an optimal one for solving the.above problem.