面向OTHR目标跟踪的多路径PHD滤波算法研究

The OTHR multi-target tracking system faces multipath, dense echo,low detection probobility and low measurement accuracy problems.Traditional OTHR tracking algorithms combine the correlation error with estimation error, so it is difficult to gain superior tracking performance in the environment with dense echo.The PHD filter has the ability of tracking multiple targets efficiently in the dense echo scenario with an integrative processing algorithm free of data association.The target number obtained by PHD filter algorithms cannot reflect the real target number due to OTHR multipath characteristics. The project intends to research on the multipath PHD filter framework in the ternary association of target state set, transformation path(mode) set and measurement set sense. As the OTHR dense echo and low measurement accuracy problems will bring high systemic computaional cost and low estimation accuracy,the multi-target multi-PHD collaborative strategy will be constructed for fast OTHR multi-target tracking with high accuracy.The aim is obtained by dividing measurement set dymatically based on the evolving characteristics of persistent target, birth target and clutter. As PHD filter can only output probability distribution information of multi-target in the state space ,and cannot provide target state and continuous target tracks information directly, the target information extraction strategy for the PHD filter will be presented by utilizing the global space-time information of the targets.

OTHR目标跟踪面临多路径、密集回波、低检测概率以及低量测精度等问题。传统OTHR跟踪算法将关联误差与估计误差耦合在一起，针对密集回波环境难以得到较好跟踪性能。PHD滤波器作为一种免数据关联的一体化处理算法，可以有效实现密集回波环境下的目标跟踪。OTHR多路径特性使得标准PHD滤波算法得到的目标个数无法反映真实目标个数。本项目拟研究目标状态集合、传播路径(模式)集合及量测集合三重关联意义下的多路径PHD滤波框架；针对OTHR密集回波以及低量测精度等问题带来的高计算量以及低估计精度问题，拟利用存活目标、新生目标以及杂波等衍化特性将量测集合动态划分，构建多PHD组协同策略，实现快速高精度的OTHR目标跟踪；针对PHD滤波器只能输出多目标在状态空间的概率分布信息，不能直接给出目标状态及连续的目标航迹信息，拟利用目标的全局时空信息，构建PHD目标信息提取策略。

针对OTHR目标跟踪面临的多路径、密集回波、低检测概率以及低量测精度等问题，本项目研究了面向OTHR目标跟踪的多路径PHD滤波算法。具体创新点如下：(1) 面向天波超视距雷达多目标跟踪的多路径PHD滤波框架；(2) 面向高精度快速跟踪的协同PHD滤波器的设计；(3) 基于PHD滤波器的全局航迹提取方法; (4)基于全局时空信息的SMC-PHD多目标跟踪算法；(5) 不均匀杂波环境及低检测概率下的改进自适应PHD滤波器；(6) 面向OTHR目标跟踪的EMDs滤波器等；通过上述研究成果，有效提升了OTHR多径目标跟踪性能。. 本项目建立了多模PHD滤波综合仿真验证平台，并进行了仿真数据和工程实测数据验证；在国内外期刊和高水平国际会议上发表论文52篇，其中SCI收录23篇，EI收录21篇；在国防工业出版社和兵器工业出版社出版专著2部(国防科技图书出版基金资助1部)，在国防工业出版社出版译著2部(均为装备科技译著出版基金资助)；授权技术发明专利4项，软件著作权10项；培养博士4人，硕士10人；获陕西省科学技术二等奖1项，国际会议Best paper 1项，国内会议优秀会议论文2项。