Design of Neural Network Adaptive Control Law for Aircraft System including Uncertainty
by
Dongho Shin
Seoul National University, Department of Mechanical and Aerospace Engineering, Seoul 151-742, KOREA
Coauthors: Youdan Kim (Seoul National University, Department of Mechanical and Aerospace Engineering, Seoul 151-742, KOREA)

Recently, high performance aircraft is designed to have more maneuverability at high angle of attack. Usually, it is very hard to obtain the accurate dynamic model for the high performance aircraft, because the aerodynamic characteristics are highly nonlinear and also include lots of uncertainties. In addition, the dynamic model may be changed much from the nominal one when the faults such as actuator fault or wing damage occur. Therefore, the performance of the control system without considering uncertainties or unmodeled dynamics caused by faults may be degraded. In this paper, to overcome these defects, neural networks based adaptive nonlinear controller is proposed making use of the backstepping technique without the assumption of two-timescale separation. Neural networks are implemented to guarantee robustness to uncertainties caused by the nonlinear aerodynamic characteristics as well as the unmodeled dynamics due to the control surface damage. The main feature of the proposed control system is that the adaptive control law is developed under the assumption that all of the nonlinear functions of the system include uncertainty, whereas most previous works assume that only some of the nonlinear functions have unknown terms. To show the effectiveness and reconfigurable ability of the proposed control law, six degree-of-freedom nonlinear model of the F-16 aircraft is used.

Date received: January 14, 2002

Copyright © 2002 by the author(s). The author(s) of this document and the organizers of the conference have granted their consent to include this abstract in Atlas Conferences Inc. Document # cago-24.