Time-Coordination Entry Guidance for Unpowered Gliding Morphing Aircrafts Using Deep Neural Networks

A time-coordination reentry guidance law using deep neural networks for Morphing aircrafts is developed in this paper. The neural network fits the mapping from states, guidance and morphing parameters to flight performances using the dataset generated by traversing bank angle profile and morphing parameters. In the guidance law, leveraging the automatic differentiation property of the neural network and Newton iteration methods, guidance and morphing parameters matching the expected range and flight time are determined. Lateral guidance is conducted based on the exponential convergence criterion for bank angle flips. Simulation results demonstrate that multiple morphing aircrafts satisfy path constraints and achieve the desired guidance accuracy, providing sufficient evidence for the effectiveness of the time-coordination guidance law.