Abstract

Fuzzy model based chaotic time series prediction has been extensively studied. However, traditional type-1 fuzzy system, whose membership functions are type-1 fuzzy set, has its limitation in handling uncertainties. Type-2 fuzzy system, whose membership functions are type-2 fuzzy set, has more adaptable parameters and design degree of freedom, making it more powerful in dealing with nonlinear and uncertain problems. In this paper, the interval type-2 fuzzy system, whose fuzzy space is divided by mesh diagonal method and consequent parameters are adapted by the forgetting factor recursive least square method while previous parameters remain unchanged, is applied to modeling and prediction of chaotic time series. Finally, the proposed scheme is tested by prediction of Mackey-Glass chaotic time series in case of noise free and noise condition, the simulation results show that the proposed scheme has a higher prediction accuracy and is powerful in handling uncertainties.

Paper Details