Aircraft Control Allocation Methods Considering Parametric Model Uncertainty

This dissertation presents the development and simulation of real-time control allocation methods which were specifically designed considering parametric model uncertainty and was performed under the NASA Modeling and Control for Agile Aircraft Development project. Previously, the NASA Learn-to-Fly project successfully showed that combining real-time control design, real-time guidance, and real-time modeling is feasible and can help eliminate the need for ground-based aerodynamic modeling in favor of simultaneous in-flight modeling and control framework development.

The real-time modeling algorithm must obtain flight data from a wide range of explanitory variables. To obtain said data, the real-time modeling procedure uses programmed test inputs to command each of the control surfaces using wide-band frequency content. These programmed test inputs, which are necessary for real time modeling, produce aircraft responses which the control system sees as disturbances and must be suppressed. This creates a fundamental conflict, good control tends to mitigate disturbances, including those due to the programmed test inputs. The basis of these control allocation methods are to separate the control design problem from the control allocation design problem. In this dissertation, the in-flight control design approach is to use loop-shaping techniques based on the rigid-body aircraft equations of motion to penalize control commands within the programmed test input frequency band. This is shown to successfully control aircraft with control commands more compatible with excitation for real-time modeling.

Due to high initial uncertainty levels, optimal control allocation methods considering parametric model uncertainty are developed and employed. Conventional control allocation methods are unable to guarantee bounds on realized moment error in the presence of bounded disturbances. Control allocation by Robust Linear Programming is developed and employed within this dissertation. The Robust Linear Programming control allocator is derived to realize a moment which is guaranteed to lie within a certain prescribed tolerance bounds in the presence of bounded disturbances.

This dissertation also considers a second optimal control allocation method, the Linear Program Direction-Preserving Allocator. Conventionally, high-efficiency control allocation does not preserve commanded moment direction, this dissertation extends control allocation by linear programming to generate the linear
programming direction preserving allocator. The linear programming direction-preserving allocator is then extended using robust control theory to develop the robust linear programming direction-preserving allocator which is shown to best maintain aircraft dynamics in the event of an actuator fault.

The suite of control and control allocation methods is simulated on linear and nonlinear aircraft models vehicles considering parametric model uncertainty. The methods show the improvement on existing algorithms when considering parametric model uncertainty.