Identifying chaotic systems using Wiener and Hammerstein cascade models (Q5936760)

It is known that in analyzing and synthesizing artificial neural networks (ANNs) one of the difficulties arises from the fact that memory and nonlinearity are intermixed. This paper deals with how to handle this situation. The authors propose two approaches, whereby the nonlinearity and the memory are in a sense separated. These two identification methods have similar functional structures, where a linear dynamic plant is cascaded with a neural network in different orders, called the Wiener and Hammerstein models, respectively.