Some new trends in identification and modeling of nonlinear dynamical systems

Abstract

This contribution attempts to give an overview of the existing framework for identification and modeling of nonlinear dynamical systems, starting from the classic work of Wiener and continuing with the recent developments in artificial neural networks. A comparison of the approaches among one another is made in terms of various practical aspects such as approximating ability, computational demand, on-line applicability, noise immunity, convergence of algorithms, and special requirements.