A direct adaptive neural network control for unmodeled nonlinear multivariable systems

This paper considers the problem of direct adaptive control for unknown nonlinear multivariable systems. Two algorithms related to neural network based adaptive control for unknown nonlinear MIMO systems are presented. The first gives a real-time or instant control signal u(k) directly while the oth...

Full description

Saved in:
Bibliographic Details
Published in:Journal of engineering and applied science (Online) 2003-10, Vol.50 (5), p.843-859
Main Authors: Al-Zohairy, T A, Wahdan, A M
Format: Article
Language:English
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This paper considers the problem of direct adaptive control for unknown nonlinear multivariable systems. Two algorithms related to neural network based adaptive control for unknown nonlinear MIMO systems are presented. The first gives a real-time or instant control signal u(k) directly while the other gives both instant value and future values of the control signal i.e., u(k), u(k + 1)... The advantages of using these two algorithms are that standard backpropagation are used to predict the output of the system and simple gradient descent rule is used to calculate u(t). A comparison with other works in which the training process involves dynamic gradient methods are given to show the importance of our work. Simulation results are presented toward the end of the paper to complement the study.
ISSN:1110-1903