基于数据驱动的机动目标跟踪  

Maneuvering Target Tracking Based on Data-driven

作　　者：马云红[1] 李欣怡 王泽霖 曹天航 王永坤 Ma Yunhong;Li Xinyi;Wang Zelin;Cao Tianhang;Wang Yongkun(Northwestern Polytechnical University,Xi’an 710072,China;AVIC Leihua Electronic Technology Research Institute,Wuxi 214063,China)

机构地区：[1]西北工业大学,陕西西安710072 [2]中国航空工业集团公司雷华电子技术研究所,江苏无锡214063

出　　处：《航空科学技术》2024年第4期97-103,共7页Aeronautical Science & Technology

基　　金：航空科学基金(2020002005300)。

摘　　要：目标跟踪是进行目标精准打击的关键,特别是对于机动目标的精确跟踪可以提高对目标状态的估计精度,从而实现目标的精确打击任务。针对机动目标跟踪问题,基于神经网络理论,本文提出了基于数据驱动的机动目标跟踪算法。通过生成大量不同机动方式运动的飞行数据,建立目标机动的数据样本库,采用基于双向输入的深度神经网络,对大量不同的机动飞行数据进行训练,生成不同机动轨迹的模型,通过轨迹序列数据与不同机动模型匹配,实现对任意机动目标的运动轨迹预测,进而实现对机动目标的跟踪。仿真结果验证了基于数据驱动进行目标跟踪算法的有效性,提供了机动目标轨迹预测的新思路和新方法,提升了轨迹预测的时效和精度。Target tracking is the key to accurate target strike,especially for maneuvering targets,which can improve the estimation accuracy of target state and achieve precise target strike tasks.This paper proposes a data-driven maneuvering target tracking algorithm based on neural network theory for the tracking problem of maneuvering targets.By generating a large amount of flight data with different maneuvering modes,a data sample library for target maneuvering is established.A deep neural network based on bidirectional input is used to train a large amount of different data,generate models of maneuvering trajectories,and match trajectory sequence data with different maneuvering models to achieve trajectory prediction for any maneuvering target,thereby achieving tracking of maneuvering targets and effectively improving the tracking accuracy of maneuvering targets.The simulation results validate the effectiveness of data-driven target tracking algorithms,provides new ideas and methods for predicting maneuvering target trajectories,and improves the timeliness and accuracy of trajectory prediction.

关 键 词：神经网络 数据驱动 机动目标 目标跟踪 轨迹预测 

分 类 号：TN953[电子电信—信号与信息处理]