ANALYSIS OF ADAPTIVE NEURAL NETWORKS FOR HELICOPTER FLIGHT CONTROL

The design of online adaptive neural networks for use in a nonlinear h elicopter flight control architecture is treated. Emphasis is given to network architecture and the effect that varying the adaptation gain has on performance. Conclusions are based on a six degree-of-freedom n onlinear evaluation model of an attack helicopter and a metric that me asures the network's ability to cancel the effect of modeling errors f or a complicated maneuver. The network is shown to provide nearly perf ect tracking in the face of significant modeling errors and, additiona lly, to cancel the model inversion error after a short initial period of learning. Furthermore, it Is shown that the performance varies grac efully and monotonically improves as the adaptation gain parameter is increased. The effect on control effort is modest and is mainly percep tible only during a short training episode that can be associated with transition from hover to forward flight.