You are here

Neural Network Control of Nonlinear Systems Using Multiple

Award Information
Agency: Department of Defense
Branch: Navy
Contract: N/A
Agency Tracking Number: 40200
Amount: $70,000.00
Phase: Phase I
Program: STTR
Solicitation Topic Code: N/A
Solicitation Number: N/A
Timeline
Solicitation Year: N/A
Award Year: 1998
Award Start Date (Proposal Award Date): N/A
Award End Date (Contract End Date): N/A
Small Business Information
2 Research Place, Suite 202
Rockville, MD 20850
United States
DUNS: N/A
HUBZone Owned: No
Woman Owned: No
Socially and Economically Disadvantaged: No
Principal Investigator
 Dr. Chiman Kwan
 () -
Business Contact
 Dr. Joseph E. Schwartz
Phone: () -
Research Institution
 University of Texas
 Frank L. Lewis 
 
7300 Jack Newell Blvd. S
Fort Worth, TX 76118
United States

 Nonprofit College or University
Abstract

Here we propose a new and general control architecture using multiple neural network models. There are three key components in our approach. First, multiple neural network models were used for feedforward control. The models store the inverse dynamics of the nonlinear plant under various operating conditions. Feedforward control basically cancels the nonlinear dynamics of the plant without affecting the closed-loop stability of the system. Second, a feedback neural network controller is used to further enhance the performance of the feedforward control. Since the cancellation of nonlinear dynamics by feedforward control may not be perfect, the residuals will be eliminated by the feedback controller. The feedback controller design is based on our on-line unsupervised learning scheme that can assure closed-loop stability. Third, a switching logic based on the principle of fuzzy logic will be used to determine when to switch from one model to another. The rule based fuzzy logic switching controller is easy to design and uses only Mach number and angle-of-attack as input variables. The logic will also decide on when to update the multiple neural net models for feedforward control. This re-learning process is necessary if there is partial or complete failure of certain actuators or sensors.

* Information listed above is at the time of submission. *

US Flag An Official Website of the United States Government