基于伯努利滤波与K秩融合的协同跟踪效能分析  

Performance Analysis of Collaborative Tracking Based on Bernoulli Filter and K-Rank Fusion

作　　者：吴俊卿 汪飞[1] 周建江[1] 韩清华 WU Junqing;WANG Fei;ZHOU Jianjiang;HAN Qinghua(Key Laboratory of Radar Imaging and Microwave Photonics Ministry of Education Nanjing University of Aeronautics and Astronautics,Nanjing 211000 China;School of Artificial Intelligence Zaozhuang University,Zaozhuang 277000 China)

机构地区：[1]南京航空航天大学雷达成像与微波光子教育部重点实验室,南京211000 [2]枣庄学院人工智能学院,山东枣庄277000

出　　处：《电光与控制》2025年第3期7-14,共8页Electronics Optics & Control

基　　金：国家自然科学基金面上项目(62271247);中国博士后科学基金面上项目(2023M731756);南京航空航天大学前瞻布局科研专项资金(ILA220171A22);枣庄学院“青檀学者”人才项目(1020701)。

摘　　要：针对多雷达协同目标跟踪的效能分析需求,首先对基于密度估计的主流融合算法建模,并结合高斯-均匀混合分布的熵误差理论分析其融合局限,然后基于序贯蒙特卡罗近似的伯努利滤波与K秩融合设计了多雷达协同跟踪方法,提出了误差减函数作为效能分析指标,用于定量分析多雷达目标跟踪的协同效能。仿真实验表明:当多雷达目标跟踪的过程噪声过大或过小时,融合跟踪性能无本质改善,表明该条件下增加雷达数量的效能低;仅当过程噪声适宜时,融合跟踪性能有明显提升,且提升幅度与雷达数量的增加呈正向递减关系,为权衡雷达节点数量开支提供了指导意义。In addressing the demand for efficiency analysis in multi-radar collaborative target tracking the mainstream fusion algorithms based on density estimation are modeled and their limitations are analyzed based on the entropy error theory of Gaussian-Uniform mixture distribution.A multi-radar collaborative tracking method based on Bernoulli filtering of sequential Monte Carlo approximation and K-rank fusion is designed.Additionally an error decreasing function is adopted as an indicator to quantitatively analyze the collaboration effectiveness of multi-radar target tracking.The simulation experiments show that:1)When the process noise of multi-radar target tracking is either too high or too low there is negligible improvement in fusion tracking performance indicating minimal efficiency gains from increasing radar quantity under such conditions;and 2)When the process noise is appropriate there is significant improvement in fusion tracking performance and the improvement showing a positive decreasing relationship with the increase of number of radars.The study provides a guidance for determining the number of radar nodes.

关 键 词：多雷达 目标跟踪 信息融合 K秩法融合 

分 类 号：TN95[电子电信—信号与信息处理]