A genetic-based method applied in fuzzy modeling

The identification of a fuzzy system model consists of two major phases: structure identification and parameter identification. The aim of the paper is to determine the main aspects involved in developing a flexible method able to learn and optimize both the structure and the parameters of a fuzzy inference system (FIS) with applications in fuzzy modeling. We propose a special kind of GA with variable length genotypes. We tried to avoid the difficult problem of designing a recombination operator for parents of different sizes because in the natural environment we usually cannot find a correspondence for it.