AUTOGENERATION OF FUZZY RULES AND MEMBERSHIP FUNCTIONS FOR FUZZY MODELING USING ROUGH SET-THEORY

The rough set theory can represent a degree of consistency between con dition and decision attributes of data pairs which do not have linguis tic information. By using this ability, a measure called occupancy deg ree is defined, which can represent the degree of consistency between premise and consequent variables in fuzzy rules describing given exper imental data pairs. A method is also proposed by which the projected d ata is partitioned on the input space, and an optimal fuzzy rule table and membership functions of input and output variables are found from data without preliminary linguistic information. The validity of the proposed method is examined by modelling data pairs which are randomly generated from a fuzzy system.