仅用距离-多普勒观测的多传感器多目标跟踪新方法

Radar systems commonly use huge antenna to obtain accurate angle measurements of target in order for realizing tracking individually, but it also makes the system very huge, complex and expensive. Is it still necessary for the node within the networked tracking system, which becomes more and more advanced, to be able to provide accurate angle measurements? This project investigates the multiple radar system, not using huge receiver antenna and nor angle measurements, to explore the mechanism and method of target tracking using only range and Doppler measurements. Integrated localization ambiguity, in other words, ghost, is the biggest challenge that this system faces to. This project starts with the research on the ghost characteristics, discovering the rules of ghost generation and evolutions, and targets at adaptive tracking methods. The new multi-sensor multi-target tracking method, which is sensitive to the ghost characteristics, based on the GM-PHD and the ghost component controlling will be explored in this project. The new state vector, which is observable in the single sensor stage, will be defined from the quantities of range, Doppler and converted Doppler. The dynamic estimation model of this new state is to be derived. Then the new theory of target tracking will be explored for single sensor scenario using only range and Doppler measurements. Also, the local track association and fusion method with adaptive de-ghosting strategies will be discussed in this project. It will not only find a new way for the development of the miniaturized and networked radars or other kinds of sensors, but also bring important revolution to the multi-sensor multi-target systems and theories and thus promote the development of estimation, tracking and radar theories. Therefore, the execution of this project is of important significances.

雷达通常采用大型接收天线以获得精确的角度测量并独立实现目标定位，但这也使得系统变得庞大、复杂且昂贵。在日益发达的网络化跟踪体系中，各节点是否仍必须提供精确的角度观测？本项目对不需大型接收天线、没有角度信息的多雷达系统进行研究，探索仅用距离-多普勒观测进行多目标跟踪的机理与方法。组合定位模糊，即鬼影问题，是该系统面临的最大挑战。本项目从鬼影的特征分析入手，揭示其形成与演化规律；研究鬼影特征敏感的引入鬼影分量控制的多传感器GM-PHD跟踪新方法；以距离-多普勒及转换多普勒为基础，定义新的单传感器可观测状态向量，推导其动态估计模型，探索仅用距离-多普勒观测的单传感器多目标跟踪新理论，并研究基于分级关联去鬼影的分布式多传感器系统目标跟踪新方法。本项目不仅为雷达等设备的轻型化、网络化发展开拓新途径，也为多传感器探测跟踪系统与理论的发展带来重大变革，推动估计、跟踪与雷达学科的发展，具有重要的学术意义。

雷达通常采用大型接收天线以获得精确的角度测量并独立实现目标定位，但这也使系统变得庞大、复杂而昂贵。在日益发达的网络化跟踪体系中，各节点是否仍必须提供精确的角度观测？本项目对不需大型接收天线、没有角度信息的多雷达系统进行研究，探索仅用距离-多普勒观测进行多目标跟踪的机理与方法。组合定位模糊，即鬼影问题，是该系统面临的最大挑战。本项目从鬼影的特征分析入手，揭示其形成与演化规律；研究鬼影特征敏感的引入鬼影分量控制的多传感器GM-PHD跟踪新方法；以距离-多普勒及转换多普勒为基础，定义新的单传感器可观测的状态向量，推导其动态估计模型，探索仅用距离-多普勒观测的单传感器多目标跟踪新理论，并研究基于分级关联去鬼影的分布式多传感器系统目标跟踪新方法。本项目不仅为雷达等设备的轻型化、网络化发展开拓新途径，也为多传感器探测跟踪系统与理论的发展带来重大变革，推动估计、跟踪与雷达学科的发展，具有重要的学术意义。基于此项目发表论文26篇，SCI收录8篇，EI收录18篇，出版译著1部(国防工业出版社，37.8万字)，获得授权专利4项。其中，5篇论文发表在JCR二区以上期刊，2篇长文发表在业内权威期刊(IEEE Transactions on Aerospace and Electronic Systems和Signal Processing)上。