基于动态模式分解的飞行轨迹预测  

Flight Trajectory Prediction Based on Dynamic Mode Decomposition

作　　者：路晶[1] 白一丹 钟一涛 陈燃 LU Jing;BAI Yidan;ZHONG Yitao;CHEN Ran(Civil Aviation Flight University of China,Guanghan 618307)

机构地区：[1]中国民用航空飞行学院,广汉618307

出　　处：《舰船电子工程》2025年第3期61-65,137,共6页Ship Electronic Engineering

基　　金：国家自然科学基金项目“学习型自治数据库系统关键技术研究”(编号:62272066)资助。

摘　　要：精准的飞行轨迹预测是保障飞行训练安全的关键举措。针对目前飞行轨迹预测精度不足的问题,基于科目训练数据的非线性相关特性,提出迭代结合HDMD和集合卡尔曼滤波(EnKF)的组合模型对飞行轨迹进行预测。首先通过HDMD算法将飞行轨迹数据升至高维空间,保留其全部特性并获取飞行轨迹的近似线性化动态模型。在此基础上,不断迭代更新EnKF对动态模型进行调整。实验结果表明,该模型能有效减少累计误差和高维噪音影响后续预测的问题,有效提高预测精度。Accurate flight trajectory prediction is a key initiative to ensure the safety of flight training.Based on the nonlinear correlation characteristics of subject training data,a combined model of HDMD and Ensemble Kalman filter(EnKF)is proposed to predict the flight trajectory iteratively.Firstly,the flight trajectory data are upgraded to high-dimensional space by HDMD algo-rithm,which retains all its characteristics and obtains an approximate linearized dynamic model of the flight trajectory.On this ba-sis,the dynamic model is adjusted by iteratively updating the EnKF.The experimental results show that the model can effectively re-duce the cumulative error and high-dimensional noise on the problem of influencing the subsequent prediction,and effectively im-prove the prediction accuracy.

关 键 词：轨迹预测 非线性系统 动态模式分解 集合卡尔曼滤波 

分 类 号：V217[航空宇航科学与技术—航空宇航推进理论与工程]