Nonlinear dynamic systems identification based on dynamic wavelet neural units.

In this paper, a dynamic wavelet network (DWN) is proposed and applied to identify black box models of the process. The well-known delta-rule is extended to the dynamic delta-rule in order to optimize wavelet network parameters. A chemical process was chosen as a realistic nonlinear system to demonstrate the identification performance. A comparison was made between the approach presented in this paper and dynamic multi layer perceptron neural networks. [ABSTRACT FROM AUTHOR]