Abstract：With the increasing attack level of hypersonic vehicles, how to accurately predict their trajectory has become critical to defense research. This article has proposed a trajectory prediction model based on parameter estimation, which combines with the Transformer deep learning model. First, this paper developed a control parameter model that affects aerodynamic vehicle maneuvers and summarized the parameters’ controlling rules under different maneuver modes. Secondly, this paper established a trajectory control parameter prediction model based on Transformer architecture and designed a neural network loss function that can balance the optimal control parameters with the physical trajectory. Finally, this article simulated trajectory data for multiple different manoeuvring modes and analyzed the changes in control parameters and trajectory data using the trajectory prediction model. This article has then obtained test results by inputting test trajectory data into the trained model. The results show that the trajectory prediction model proposed in this article performs precise prediction of hypersonic vehicle trajectories under different manoeuvring modes.
狄子琦, 王翔宇, 吴双, 周宇 . 基于Transformer架构的高超声速飞行器轨迹生成与预测算法[J]. 空天防御, 2023, 6(4): 35-41.
DI Ziqi, WANG Xiangyu, WU Shuang, ZHOU Yu. An Algorithm for Trajectory Generation and Prediction of Hypersonic Vehicle Based on Transformer Architecture. Air & Space Defense, 2023, 6(4): 35-41.
