MEMORY NEURON NETWORKS FOR IDENTIFICATION AND CONTROL OF DYNAMICAL-SYSTEMS

This paper discusses Memory Neuron Networks as models for identificati on and adaptive control of nonlinear dynamical systems. These are a cl ass of recurrent networks obtained by adding trainable temporal elemen ts to feed-forward networks that makes the output history-sensitive. B y virtue of this capability, these networks can identify dynamical sys tems without having to be explicitly fed with past inputs and outputs. Thus, they can identify systems whose order is unknown or systems wit h unknown delay. It is argued that for satisfactory modeling of dynami cal systems, neural networks should be endowed with such internal memo ry. The paper presents a preliminary analysis of the learning algorith m, providing theoretical justification for the identification method. Methods for adaptive control of nonlinear systems using these networks are presented. Through extensive simulations, these models are shown to be effective both for identification and model reference adaptive c ontrol of nonlinear systems.