INTEGRATED DESIGN OF AN ADAPTIVE NEUROCONTROLLER FOR A 2-D AEROELASTIC SYSTEM

The design of control configured structures has been considered in a n umber of recent studies. Both active and passive measures for structur al vibration control have been examined in this context. The present p aper addresses issues related to the use of neural network based contr ol systems in such applications. A simplified 2-D representation of an aeroelastic system, consisting of an airfoil with a trailing-edge fla p, comprises the test bed for the present study. With a proper selecti on of structural spring characteristics, and choice of unsteady aerody namic forces and moments, the system provides a rudimentary 2-D model of a helicopter rotor blade that includes both structural and aerodyna mic nonlinearities. The integrated optimal design of the plant and its control system for optimized response under disturbance loading is th e principle objective of the design exercise. The focus of the paper i s three-fold - it establishes the justification for replacing traditio nal control systems with neurocontrollers in such problems, examines i ssues related to an integrated structural-control design strategy, and discusses a detailed implementation of the approach in a linearized 2 -D aeroelastic system. The design problem contains multiple relative o ptima, and the use of a genetic algorithm (GA) based optimization proc edure is shown to be an effective tool to locate the optimal design. R esults from numerical experiments are presented in support of the prop osed design approach.