Adaptive Neural Networks-Based Dynamic Inversion Applied to Reconfigurable Flight Control and Envelope Protection Under Icing Conditions

Aircraft icing can result in degradation of the aerodynamic characteristics and reduction of the control effectiveness, which would pose serious threats to flight safety. Reconfigurable flight control and envelope protection of iced aircraft have become an effective solution to ensure flight safety in icing encounters. In this study, the dynamic model of the iced aircraft was established, and high precision numerical simulation method and wind tunnel virtual flight experiment were applied to obtain the icing aerodynamic database. Furthermore, the reconfigurable flight control law was designed by using the adaptive neural networks based dynamic inversion (ANN-DI) control method. Simulation results demonstrate that the control method behaves well tracking performance and strong robustness in the presence of modeling errors and control surface damage. After that, an icing envelope protection system was designed based on control limiting strategy and the ANN-DI control was applied to calculate the control surface deflection limits based on the limit values of the key flight safety parameters. Finally, the designed icing envelope protection system has been verified through simulation in two autopilot modes under icing conditions. The simulation results obtained here show that the system could keep the key flight safety parameters such as the flight speed, the angle of attack (AOA), the side slip angle, and bank angle within the flight safe region under icing conditions. The method proposed in this study is expected to provide flight safety measures for in-flight icing.