A new approach for neural control of nonlinear discrete dynamic systems

Authors:
Jose I. Canelon;Leang S. Shieh;Nicolaos B. Karayiannis
Affiliations:
Department of Electrical and Computer Engineering, University of Houston, Houston, TX;Department of Electrical and Computer Engineering, University of Houston, Houston, TX;Department of Electrical and Computer Engineering, University of Houston, Houston, TX
Venue:
Information Sciences: an International Journal
Year:
2005

Citing 9
Cited 2

Multilayer feedforward networks are universal approximators

Neural Networks
Learning internal representations by error propagation

Parallel distributed processing: explorations in the microstructure of cognition, vol. 1
First- and second-order methods for learning: between steepest descent and Newton's method

Neural Computation
Neural networks for control systems: a survey

Automatica (Journal of IFAC)
Original Contribution: A scaled conjugate gradient algorithm for fast supervised learning

Neural Networks
Discrete-time control systems (2nd ed.)

Discrete-time control systems (2nd ed.)
Application of Neural Networks to Adaptive Control of Nonlinear Systems

Application of Neural Networks to Adaptive Control of Nonlinear Systems
Stabilizing controller design for uncertain nonlinear systems using fuzzy models

IEEE Transactions on Fuzzy Systems
Control of nonlinear dynamical systems using neural networks. II. Observability, identification, and control

IEEE Transactions on Neural Networks

Indirect adaptive control of nonlinear dynamic systems using self recurrent wavelet neural networks via adaptive learning rates

Information Sciences: an International Journal
A fuzzy Actor-Critic reinforcement learning network

Information Sciences: an International Journal

Quantified Score

Hi-index	0.07

Visualization

Abstract

This paper proposes a new approach to control nonlinear discrete dynamic systems, which relies on the identification of a discrete model of the system by a neural network. A locally equivalent optimal linear model is obtained from the neural network model at every operating point the system goes through during the control task. Based on the linear model, a linear state-space control design technique can be used to design a local control action to be applied at that particular operating point. The design procedure is repeated at every operating point of the system during the control task. The proposed approach was applied in three examples, which involved a linear and two nonlinear discrete single-input single-output (SISO) systems. In the first two examples, pole placement was chosen as the linear design technique. This method led to satisfactory tracking of the reference input but a steady-state error was present due to modeling inaccuracies. In the third example, the linear design technique involved an optimal control scheme with an integrator. This method led to satisfactory tracking of the reference input with zero steady-state error.