Evolutionary interval type-2 fuzzy systems using continuous ant colony optimization algorithms

This paper proposes the optimization of interval type-2 fuzzy systems (IT2FSs) through continuous ant colony optimization (ACO) algorithm. The optimized IT2FS consists of Mamdani-type fuzzy rules, where the antecedent and consequent parts use interval type-2 Gaussian fuzzy sets with uncertain means. Given the structure of an IT2FS, this paper proposes the optimization of all the free parameters in it through two types of continuous ACO algorithms. The first one is ant colony optimization in real space, where a colony of solution vectors is created with each vector comprising all of the free parameters in an IT2FS. The second one is cooperative continuous ACO (CCACO), where multiple colonies are created with each solution vector in a colony comprising only the free parameters in a single rule. Simulations are presented to the show the performance of the continuous ACO algorithms for IT2FS design with comparisons with different type-1 and type-2 fuzzy systems.