A gamma memory neural network for system identification

Author

Motter, Mark A. ; Principe, Jose C.

Author_Institution

Facility Autom. Controls, NASA Langley Res. Center, Hampton, VA, USA

Volume

5

fYear

1994

fDate

27 Jun-2 Jul 1994

Firstpage

3232

Abstract

A gamma neural network topology is investigated for a system identification application. A discrete gamma memory structure is used in the input layer, providing delayed values of both the control inputs and the network output to the input layer. The discrete gamma memory structure implements a tapped dispersive delay line, with the amount of dispersion regulated by a single, adaptable parameter. The network is trained using static backpropagation, but captures significant features of the system dynamics. The system dynamics identified with the network are the Mach number dynamics of the 16 Foot Transonic Tunnel at NASA Langley Research Center, Hampton, Virginia. The training data spans an operating range of Mach numbers from 0.4 to 1.3

Keywords

identification; learning (artificial intelligence); neural nets; 16 Foot Transonic Tunnel; Mach number dynamics; NASA Langley Research Center; discrete gamma memory structure; gamma memory neural network; static backpropagation; system dynamics; system identification; tapped dispersive delay line; Circuit testing; Control systems; Delay lines; Dispersion; Fans; Foot; NASA; Neural networks; System identification; System testing;

fLanguage

English

Publisher

ieee

Conference_Titel

Neural Networks, 1994. IEEE World Congress on Computational Intelligence., 1994 IEEE International Conference on

Conference_Location

Orlando, FL

Print_ISBN

0-7803-1901-X

Type

conf

DOI

10.1109/ICNN.1994.374753

Filename

374753