Identification of nonlinear dynamic systems using a rapid neural network

Abstract

This paper presents the application of a rapid neural network for identification of unknown nonlinear dynamic systems when the inputs and outputs are accessible for measurements. The learning algorithm of the rapid neural network does not need an iterative procedure as in most learning algorithms such as the well known back-propagation algorithm. It is able to achieve a solution in one time training. The algorithm can be applied off-line or on-line. The algorithm is implemented and applied for identification of various types of nonlinear dynamic systems. Simulation results show excellent performance of the rapid neural network for identification of SISO and MIMO nonlinear dynamic systems.