Abstract:
A novel real-time guidance method based on deep learning is proposed to deal with the launch vehicle under thrust drop. The trajectory optimization of the LV is introduce...Show MoreMetadata
Abstract:
A novel real-time guidance method based on deep learning is proposed to deal with the launch vehicle under thrust drop. The trajectory optimization of the LV is introduced first, and the distinctions of the optimization objective and constraints are proposed when thrust drop happens. Then, design of the deep neural network is motivated by the aim of the trajectory optimization, and a real-time close- loop guidance method is proposed with the flight states introduced into the guidance loop. Next, generous optimal trajectories of all-around fault modes are generated as the sample set to train the DNN. Avoiding handling the atmosphere which is hard to express analytically, the guidance method based on deep learning can accommodate thrust drop even in atmosphere. In the end, the effectiveness and performance of the proposed method are evaluated in run time, guidance error and robustness.
Published in: 2021 China Automation Congress (CAC)
Date of Conference: 22-24 October 2021
Date Added to IEEE Xplore: 14 March 2022
ISBN Information: