A Study on the Evolutionary Adaptive Defuzzification Methods in Fuzzy Modeling

Article type: Research Article

Authors: Cordón, Oscar^a | Herrera, Francisco^a | Márquez, Francisco Alfredo^b | Peregrín, Antonio^b

Affiliations: [a] Dept. of Computer Science and Artificial Intelligence, University of Granada, 18071 - Granada, Spain. ocordon@decsai.ugr.es, herrera@decsai.ugr.es | [b] Dept. of Electronic Engineering, Computer Systems and Automatics, University of Huelva, Huelva, Spain. alfredo.marquez@diesia.uhu.es, peregrin@diesia.uhu.es

Abstract: Evolutionary Adaptive Defuzzification Methods are a kind of defuzzification methods based on using a parametrical defuzzification expression tuned with evolutionary algorithms. Their goal is to increase the accuracy of the fuzzy system without loosing its interpretability. They induce a kind of rule cooperation in the defuzzification interface. This paper deals with Evolutionary Adaptive Defuzzification Methods. We study their common general expression, the different defuzzification methods that can be obtained from it, their interpretation, and their accuracy. We consider two applications in order to analyze their accuracy in practice. We get some useful results for practical fuzzy systems designed by means of this kind of Intelligent Hybrid System.

Keywords: fuzzy modelling, adaptive defuzzification methods, evolutionary algorithms

DOI: 10.3233/HIS-2004-11-206

Journal: International Journal of Hybrid Intelligent Systems, vol. 1, no. 1-2, pp. 36-48, 2004