Identification of Nonlinear Dynamical Systems using A Higher Order Multi-Layer Neural Networks

A new neural network architecture, call a higher order multi-layer neural networks (HOMLNN) is presented. The architecture of an HOMLNN is a modified model of the evolved functional neural network (EFNN)with a hidden layer which is composed of self-evolve neurons and additional multiplication inputs between conventional inputs and self-evolve neurons. The authors drive a generalized dynamic backpropagation algorithm and show a new approach to the identification of dynamical systems by means of HOMLNN. Experiment result showed that the method is effective for the identification of dynamical systems