Polár-vágat alapú fuzzy halmaz-interpoláció

Systems working with fuzzy logic apply a rule based r asoning process. In several practical applicatio ns the rules belonging to the knowledge base do not cover all the possible input values. In such cases the cl assical fuzzy inference methods (e.g. Compositional Rule of Infer ence) must be substituted by approximate techniques . The methods following the ideas of the generalized meth odology of fuzzy rule interpolation (GM) [1] determine the approximated conclusion in two steps. First a new r ule is interpolated in the position of the observat ion and next the result is calculated by firing this rule. In th is paper we propose a new fuzzy set interpolation t echnique called FEAT-p applicable in the first step of the GM for t he determination of the antecedent and consequent p art of the new rule. Its basic concepts are the virtual lingui stic term shifting and the polar cut. As advantages can be mentioned the low computational complexity and the ability to handle even cases when one or more of th e uzzy sets are subnormal.