Share Email Print
cover

Proceedings Paper

Predictive neuro control of vibration in smart structures
Author(s): Lawrence E. Pado; Rajendra R. Damle
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

Neural network based predictive control for random vibration suppression has been demonstrated on a cantilevered beam with bonded piezoelectric actuators and sensors. This real time system is run on a 60 MHz Pentium processor and is considered a stepping stone to both adaptive flutter suppression and buffet load alleviation in advanced aerospace vehicle. An extended neural control system using Intel/Nestor's ETANN analog neural network chip is discussed. Generalized neural predictive control uses a neural network based model of a system to make predictions about the effect of future control signals on the response of the system. These predictions can be used with a tailored performance index to determine the optimal control signal for the modeled system. A comparison of this approach with both PID and pole placement control methods shows ease of implementation comparable to that of the PID controller with the approximate performance of the pole placement method. The advantages of neural control over conventional control techniques include a simpler and more cost effective design methodology as well as the capability to learn on-line the time varying nature of a system due to wear, loss of actuators, or other causes.

Paper Details

Date Published: 31 May 1996
PDF: 9 pages
Proc. SPIE 2715, Smart Structures and Materials 1996: Mathematics and Control in Smart Structures, (31 May 1996); doi: 10.1117/12.240831
Show Author Affiliations
Lawrence E. Pado, McDonnell Douglas Aerospace (United States)
Rajendra R. Damle, Univ. of Missouri/Rolla (United States)


Published in SPIE Proceedings Vol. 2715:
Smart Structures and Materials 1996: Mathematics and Control in Smart Structures
Vasundara V. Varadan; Jagdish Chandra, Editor(s)

© SPIE. Terms of Use
Back to Top