An integrated interpretable control effectiveness learning and nonlinear control allocation methodology for overactuated aircrafts

The paper presents a novel methodology for control allocation in overactuated aircrafts that integrates interpretable control effectiveness learning with nonlinear control techniques. Utilizing Sparse Identification of Nonlinear Dynamics, it constructs a compact and interpretable analytical model of control effectiveness from flight data, which allows for efficient computation in real-time without the need for onboard models. This approach demonstrates comparable accuracy to traditional nonlinear models while significantly lowering computational costs, thereby enhancing performance and robustness in dynamic flight conditions and facilitating fault diagnosis and adaptation.