This paper describes two basic structures for identifying chaotic systems b
ased on the Wiener and Hammerstein cascade models, in which three-layer fee
dforward artificial neural network is employed as the nonlinear static subs
ystem and a simple linear plant is used as the dynamic subsystem. Through t
raining of the neural network and choosing an appropriate linear subsystem,
various chaotic systems can be well identified by these two basic structur
es. Computer simulation results on Henon and Lozi systems are presented to
demonstrate the effectiveness of these proposed structures. It is also show
n that two chaotic systems whose outputs are different can actually exhibit
similar chaotic attractors. (C) 2001 Elsevier Science Ltd. All rights rese
rved.