Suboptimal intelligent control: Agile aircraft application

Near-optimal control of an aircraft using artificial neural networks is discussed with the optimal control problem solved off-line for numerous initial and final conditions using standard techniques. The obtained trajectories are then used to train a neural network controller. The concept is applied to minimum-time control and quadratic-performance optimal control of the longitudinal motion of a super-agile aircraft. Full nonlinear dynamics with control (magnitude and rate) constraints as well as state constraints are considered. Suboptimal feedback synthesized by a neural network is tested on a simulated rapid maneuver with extremely large changes in pitch angle and angle of attack.