NEURAL-NETWORK-BASED CONTROLLER FOR NONLINEAR AEROELASTIC SYSTEM

Attenuation of vibratory response is an important design consideration in many aeroelastic systems, and active methods of vibration reductio n have been studied extensively in this context, Synthesis of active c ontrollers requires that a good analytical model of the system be avai lable, In those problems in which the aeroelastic system is inherently nonlinear, a robust control scheme is difficult to implement, particu larly in the presence of large uncertainties in the model, The use of artificial neural networks, with on-line learning capabilities, is exp lored as an approach for developing robust control strategies for such problems. In particular, the use of neural networks to mimic the beha vior of a linear quadratic Gaussian controller that is applicable to n onlinear systems is presented. The helicopter rotor blade is a classic example of an aeroelastic system in which vibration reduction is an o verriding concern, and in which the plant is both nonlinear and contai ns uncertainties. A simplified two-dimensional representation of this aeroelastic system, consisting of an airfoil with a trailing-edge cont rol. flap, is considered as a test case in the present work; both stru ctural and aerodynamic nonlinearities are included in the problem.