Gain-Scheduled Model Reference Adaptive Control for a Guided Projectile

This paper documents current research on a gain-scheduled model reference adaptive control approach for projectiles operating within an extended flight envelope. Baseline control for the projectile is attained using a three-loop autopilot gain-scheduled across angle-of-attack and airspeed. A model reference adaptive control architecture is augmented to the baseline gain-scheduled controller to suppress the effect of system uncertainties. This methodology is implemented on a generic tail-controlled guided projectile to illustrate the approach. The baseline three-loop autopilot is tuned using a low-fidelity aerodynamic model of the projectile. Differences between this low-fidelity model used for controller design and a separate, highfidelity "truth" model used for simulation result in degraded performance of the baseline controller, illustrating the challenges of control design using limited system information. The model reference adaptive control augmentation is shown to suppress much of the effect of this model mismatch and recover desired performance across the flight envelope.