Nonlinear MPC for Stabilizing the Longitudinal Dynamics of a Highly Maneuverable Aircraft

This paper investigates the application of nonlinear model predictive control (MPC) for stabilizing the longitudinal dynamics of a highly unstable aircraft model. Unlike conventional MPC applications in aerospace that typically linearize the aircraft dynamics and focus on outer loop autopilot control, the nonlinear system formulation of an aircraft including a detailed aero-data model is used in the optimization problem for inner loop stabilization, enhancing control accuracy and robustness across the entire operating flight envelope. The implementation with a gradient-based MPC toolbox allows for a fast optimization by utilizing the gradients of the system dynamics as well as the aerodynamic coefficients, leading to a fast and efficient application that can accurately predict the future system behavior. By additionally considering current pilot stick deflections in the optimization problem, the control system dynamically adjusts to the pilot's commands, creating a responsive and adaptive flight control system relevant for real-flight operations. The approach is validated for the ADMIRE model (“Aero-Data Model In a Research Environment”), a high-fidelity aircraft benchmark model that includes transport delay of signals, actuator dynamics, and uncertainties. Simulations demonstrate the improved performance compared to a conventional differential PI controller. Additionally, the effectiveness and inherent robustness of the MPC strategy is highlighted by stability evaluations across the aircraft's flight envelope and under aerodynamic uncertainties. These evaluations are a key aspect for the flight clearance process, ensuring the control system's reliability under diverse and challenging flight conditions.