Abstract
Fuzzy reasoning methods are generally classiied into two approaches : direct approach and truth space approach. Several researches on the mathematical relationships between the direct approach and the truth space approach have been reported. There has been, however, no research which discusses the utility of the direct approach and the truth space approach.The authors have proposed three types of fuzzy neural networks (FNNs) celled Type I, II and III. The FNNs can identify the fuzzy rules and tune the membership functions of fuzzy reasoning atuomatically utilizing the learning capability of neural network. The TypeIII based on the truth space approach, especially, can obtain the linguistic fuzzy rules. But this type of FNN has some difficulties in understanding the fuzzy rules.This paper presents new types of FNNs called TypeIV and V based on the truth space approach. The fuzzy rules identified with the TypeIV can be comprehended more clearly than those with the TypeIII and the fuzzy rules of the TypeV can be expressed more flexibly than those of the TypeIII and IV. This paper also describes a method to label the fuzzy variables in the consequences with the linguistic truth values of the obtained fuzzy rules. The feasibility of the new FNNs are examined using simple numerical data. The results show that the truth space approach makes the fuzzy rules easy to understand.
