针对模糊推理的FMT(fuzzy modus tollens)问题,作为三I*算法的推广与改进形式,研究了FMT-泛三I*算法。首先,分析了FMT-泛三I*算法的属性,提出了该算法的基本原则,改进了之前三I*算法的原则。其次,面向R-蕴涵算子,建立了FMT-泛三I*算法的统一形式的解,同时针对几类经典的R-蕴涵算子,分别获得了具体情形下的优化解。最后,证明了FMT-泛三I*算法的置换还原性,获得了良好效果。 Aiming at the fuzzy modus tollens (FMT) problem of fuzzy inference, the FMT-pan-triple I * algorithm is studied as a generalization and improvement of the triple I * algorithm. Firstly, the properties of FMT-pan-three I * algorithm are analyzed, the basic principle of the algorithm is proposed, and the principle of the former three I * algorithm is improved. Secondly, a generalized form of the FMT-pan-triple I * algorithm is developed for R-implication operators. For the classical R-implication operators, the optimal solutions are obtained for each case. Finally, we proved that the FMT-pan-triple I * algorithm has good reduction effect.