基于随机有限集的机载多普勒雷达多目标跟踪方法研究

The airborne radar can cooperate with the ground radar and realize the compensation of the blind zone. This will effectively respond to the low-altitude target’s penetration. However, owing to airborne radar working in the look-down mode and the existence of many blind zones (e. g., Doppler blind zone), target tracking usually is faced with the following problems, for instance, clutter is dense, multiple targets should be tracked, the detection points may be easily missing, and the batch number may be reinitiated. These problems are difficult to resolve by using the classical multi-target tracking (MTT) algorithms based on data associations. The emerging random finite set (RFS)-based multi-target tracking (RFS-MTT) algorithms avoid complex data association, and increasingly importance has been attached to them by scholars both at home and abroad. Based on the RFS theory, this project makes full use of the Doppler information from the Doppler radar, and attempt to study the MTT approaches in dense clutter for airborne Doppler radars. The main contents include as follows. First, the RFS-MTT algorithm with incorporating Doppler information is studied to impress clutter and improve the MTT performance. Second, the RFS-MTT algorithm with the Doppler and minimum detectable velocity information is studied to reduce the influence of missing points caused by the Doppler blind zone. Finally, the general RFS-MTT algorithm with labels is studied to provide track labels and overcome the problem of reinitiation batch number. The research findings could prompt the RFS’s booming development, and provide a novel solution for target tracking with airborne Doppler radars. Thus, they have significant both academic worth and engineering application value.

机载雷达可与地面雷达协同补盲，有效应对低空目标突防。然而，由于机载雷达下视工作以及多种盲区（如多普勒盲区）的存在，目标跟踪常面临杂波密、目标多、易掉点、重起批等问题。这些问题难以用常规的基于数据关联的多目标跟踪（MTT）算法解决，新兴的随机有限集（RFS）多目标跟踪（RFS-MTT）算法避免了复杂的数据关联运算，正受到国内外学者高度重视。本项目基于RFS理论，充分利用机载多普勒雷达的多普勒信息，探索密集杂波下机载多普勒雷达MTT方法，主要研究内容有：为抑制杂波提高MTT性能，研究并入多普勒信息的RFS-MTT算法；为降低多普勒盲区造成的掉点影响，研究并入多普勒和最小可检测速度信息的RFS-MTT算法；为提供航迹标签并克服重起批问题，研究一般的带标签的RFS-MTT算法。研究成果将推动RFS的蓬勃发展，将为机载多普勒雷达目标跟踪难题提供新的解决思路，具有重要的学术价值和工程应用价值。

针对机载雷达跟踪低空目标面临的“杂波密、目标多、易掉点、重起批”等现实难题，项目基于最先进的适用于多目标跟踪的随机有限集（RFS）理论，首先利用多普勒量测抑制杂波，提出了带多普勒量测的多伯努利（MB）滤波器的两种实现方法（即高斯混合（GM）和序贯蒙特卡洛（SMC）实现）；接着利用最小可检测速度（MDV）信息缓解多普勒盲区（DBZ）导致的易掉点问题，提出了带多普勒和MDV的MB滤波器的两种实现方法；然后提出了带多普勒、标签和自适应航迹起始（ATI）的MB滤波器的两种实现方法，克服标准RFS滤波器不能输出航迹标签且局限于固定位置起始的问题；最后集成上述成果，提出了带多普勒、MDV、标签和ATI的MB滤波器的两种实现方法（即GM/SMC-MBwD&MDV&L&ATI滤波器），系统解决前述“杂波密、目标多、易掉点、重起批”难题。在此基础上，将上述策略应用于概率假设密度（PHD）和带势PHD（CPHD）滤波器，开发了带多普勒、MDV、标签和ATI的PHD和CPHD滤波器的GM实现方法，并与提出的GM/SMC-MBwD&MDV&L&ATI滤波器进行了对比分析，验证了所提滤波器的优势。在完成上述既定研究内容后，进一步开展了拓展深化研究。针对多目标机动使得易掉点重起批更加频繁的问题，严格推导了多模型（MM）广义标签多伯努利（GLMB）滤波器，分别提出了基于MM-PHD和MM-GLMB滤波器的DBZ下多机动目标跟踪方法。此外，为进一步抑制DBZ的影响，利用DBZ受目标与机载雷达之间相对几何和相对速度影响的物理机理，分别提出了单预警机飞行航线实时规划方法和多预警机数据融合方法。仿真验证了上述方法的可行性和优越性，正在开展实测数据检验。项目资助期内，负责人荣获全军优秀博士论文，入选中国科协军事科技领域青年人才托举工程培养对象，已出版著作2部，即将出版1部，刊出或录用论文13篇，在审SCI论文6篇，申请发明专利4项。研究成果推动了RFS理论的发展，为机载雷达目标跟踪难题提供了新的解决途径，具有重要的学术价值和工程应用价值。