Applications of Fuzzy Pattern Recognition to Function Mapping and Controller Design

摘要

A novel mathematical tool to construct fuzzy mapping from multiple inputs to single output, for a given finite set of training data, is introduced. In contrast to available heuristic partition methods, the algorithm uses fuzzy pattern recognition theory to provide a systematic approach for the selection of discrete fuzzy bases functions from which suitable, continuous fuzzy bases functions may be obtained. Applications of the proposed method to system identification and controller design are presented.