ABSTRACT
This chapter discusses the fundamental problem of non-linear dynamic systems identification using artificial neural networks and provides some computer simulations. ANNs have been applied to regulation/tracking problems and optimal control. The chapter considers identification of nonlinear systems using multi-layer perception (MLP) neural networks and also considers the controllability of the system. The application of MLP neural networks to nonlinear system identification is restricted to systems which have an asymptotically stable equilibrium point. To use the MLP neural network for system identification, the system to be identified and excitation should meet certain conditions, otherwise the identification performance will be very poor. The input-output difference equation model for discrete nonlinear systems was proposed by I. J. Leontaritis and S. A. Billings. As the foundation of the nonlinear system identification scheme is a static nonlinear mapping approximation, the chapter also discusses the identification of static nonlinear mappings.
