针对在基于合成推理规则(CRI)的近似推理研究中,模糊蕴涵算子的选择这一核心问题,对采用t-范数与t-余范数以及否定算子所构造的模糊蕴涵算子进行归纳,分为五类(t,S,R,QL和D蕴涵),为构造新的算子提供了强有力的依据;采用绝大多数模糊规则芯片中使用的CRI方法,依据广义取式推理与广义拒式推理中的12条模糊推理准则对分类后的模糊蕴涵算子进行实际的逻辑推演,通过对逻辑推演中实例数据和推演结果的分析,给出选择模糊蕴涵算子的参考依据.
References is proposed to select fuzzy implication operators which are the one of the core issues in approximate reasoning. The five classifications (t, S, R, QL and D-implication) of the fuzzy implication function were discussed, which usually are constructed from t-norms, t-conorms and negation operators. Further, in accordance with the generalized modus ponens (GMP) rules and the generalized modus tollens (GMT) rules, the quantitative inference results for these classification operators are given by using compositional rule of inference (CRI).