Minimal Modeling Adaptive Control of the NASA Generic Transport Model with Unknown Control-Surface Faults

Sudden or gradual faults in sensors, actuators, and processors can have catastrophic consequences for plants that depend on control technology. This paper applies adaptive control to an aircraft with unknown control-surface faults and investigates the robustness of the controller to uncertainties in the aircraft dynamics. Control-surface faults include magnitude and rate saturation, deadzone, and jams in the ailerons and rudder. Retrospective cost adaptive control (RCAC) is applied in both centralized and constrained architectures with a single tuning for each architecture in order to compare the range of unknown faults that RCAC can compensate for through on-line adaptation. The results show that the centralized controller architecture can generally compensate for a larger range of fault parameters. We then characterize the robustness of the centralized controller architecture with respect to erroneous prior modeling information.