Comparison of operations used for fuzzy modeling

In this paper, fuzzy if ...then rules are used for modeling. The model is considered as a multi-dimensional function, which must approximate a real object. Special attention is devoted here to the if part of the rules. Conclusion of each rule is assumed as fuzzy singleton, as in many practical cases. The author shows that the Mamdani approach is not the best solution. There are some other operations than the minimum used for "and" which give smaller modeling of the error. In the paper, some possible interpretations of connective "and" are considered and compared from modeling the error point of view.