Deep Neural Network-Based Penetration Trajectory Generation for Hypersonic Gliding Vehicles Encountering Two Interceptors

A deep learning-based approach is developed to address the penetration problem for the hypersonic gliding vehicle (HGV) in this paper. The dynamics of HGV and the penetration problem are formulated. Then, a recently proposed second-order cone programming (SOCP) method is utilized to solve penetration trajectory optimization problems. Furthermore, the generated trajectories are used to train the deep neural network (DNN), which is then applied as the command generator. The DNN-based controllers are trained offline and act as real-time controllers online. Simulations are carried out to verify the effectiveness and real-time performance of the DNN-driven scheme.