A study of neural networks for flight control

The application of two neural network control paradigms for inner loop flight control is examined. The task of each is control of the longitudinal axis of a VISTA/F-16 aircraft model. Each design makes use of a reference model which has been developed a priori to provide the desired performance. Designs are carried out using a high fidelity nonlinear simulation. The ability of each controller to adapt its performance to variations in the plant dynamics are assessed, as well as the ability of the controllers to generalize to situations for which they were not trained.