摘要
提高轨迹预报精度是反高速机动目标面临的难点之一。本文基于目标运动特性分析,构建跟踪运动模型,采用交互多模型滤波算法完成多模型交互,局部滤波器采用自适应高阶容积卡尔曼滤波算法,通过与强跟踪滤波算法相结合,提高传统的估计精度,进而提高算法的鲁棒性;采用广义回归神经网络设计轨迹预报算法,并在预报过程中引入预报修正量对误差进行修正,通过样本学习提高长时间预报精准度,仿真结果表明,设计的跟踪预报算法在预报精度上较常规算法有较大幅度提升。
Improving the accuracy of trajectory prediction is one of the difficulties faced by anti-high-speed strong maneuvering targets.The tracking motion model is established,which is based on the analysis of typical target motion characteristics in this paper.And the interaction multi-model(IMM)filtering algo-rithm is adopted to complete multi-model interaction.The local filter adopts adaptive high-order cubature Kalman filtering(HCKF)algorithm,and the strong tracking filtering(STF)algorithm is combined with HCKF algorithm to improve the estimation accuracy of traditional CKF and the robustness of the algorithm.The prediction extrapolation method is designed,which is based on the generalized regression neural network(GRNN),and the prediction correction is introduced in the prediction process to correct the error,and the accuracy of long-term prediction is improved by sample learning.The simulation results show that the fore-cast accuracy of the proposed tracking prediction algorithm is much improved by the compared conventional algorithm.
作者
李瑞康
廖欣
张诞
唐胜景
Li Ruikang;Liao Xin;Zhang Dan;Tang Shengjing(Shanghai Electro-Mechanical Engineering Institute,Shanghai 201109,China;Northwestern Polytechnical University,Xi’an 710072,China;Beijing Institute of Technology,School of Astronautics,Beijing 100081,China)
出处
《航天控制》
CSCD
北大核心
2023年第6期3-10,共8页
Aerospace Control
基金
上海航天科技创新基金项目(SAST2017-11)。
关键词
轨迹预报
高速强机动
交互多模型
广义回归神经网络
Trajectory prediction
High-speed and strong-maneuvering
Multi-model interaction
Gener-alized regression neural network(GRNN)
