Identification of nonlinear systems by fuzzy models

Abstract

The concept of sampled probability distributions is introduced. A new formulation of the unified identification problem of quasi-linear fuzzy models (QLFMs) and quasi-nonlinear fuzzy models (D. Filev, 1990, 1991) that considers simultaneously the structure and parameter identification is proposed. A learning algorithm realizing structure and parameter identification of QLFMs is proposed.<>