Fuzzy reasoning method for smooth interpolation

Abstract

The paper describes a fuzzy reasoning method whose result belongs to the C/sup 1/ class. The proposed fuzzy reasoning method refers to interpolation by man. We suppose that humans consider two directions while interpolating. A direction of fluctuation at a given point is one, and a direction toward neighboring data is the other. Both directions are reflected in the consequent part of the fuzzy rule. It is not necessary to give them as knowledge, these are only decided as given data pairs. Even if new data is given, if is not necessary to increase a rule. The reasoning result can correspond to new data without increasing a rule. The effectiveness of the present method as verified by applications to practical data and by computer simulations.