Problems of identification of cloud-based fuzzy evolving systems

Abstract

In this paper some problems that arise in identification of nonlinear systems described by the cloud-based fuzzy rule-based model are shown. These models do not assume fixed partitioning of the space of antecedent variables. The Mahalanobis distance among the data samples is proposed for local density calculation which is more suitable when the data are scattered around the input-output surface. The identification algorithms are given in a recursive form which is necessary for the implementation of an evolving system. The proposed algorithms are illustrated on a simple simulation model of a static system.