• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

J4 ›› 2016, Vol. 38 ›› Issue (04): 833-838.

• 论文 • 上一篇    

天基光学观测低轨多目标跟踪的多模型CPHD滤波方法

李冬,玄志武   

  1. (91550部队94分队,辽宁 大连 116023)
  • 收稿日期:2015-02-16 修回日期:2015-08-11 出版日期:2016-04-25 发布日期:2016-04-25

Multimodel CPHD filtering for LEO multitarget
tracking with spacebased optical observation       

LI Dong,XUAN Zhiwu   

  1. (Unit 94,Troop 91550,Dalian 116023,China)
  • Received:2015-02-16 Revised:2015-08-11 Online:2016-04-25 Published:2016-04-25

摘要:

低轨多目标跟踪是天基光学系统信息处理需要解决的重要问题之一。提出了一种基于多模型势概率假设密度(CPHD)滤波的跟踪方法,建立了描述低轨目标运动的常轴向力模型和二体力学模型,给出了天基测量模型,将低轨目标的运动模式和运动状态组合成扩展状态,利用CPHD滤波递推扩展状态的验后概率假设密度(PHD)和目标数量的验后概率密度,能够同时得到目标状态和目标数量的估计。仿真结果表明,多模型CPHD滤波对目标数量和目标状态的估计精度相对多模型PHD滤波和单模型CPHD滤波有显著提高。

关键词: 天基光学观测, 多模型, 势概率假设密度滤波, 低轨, 多目标跟踪

Abstract:

Lowearthorbit (LEO) multitarget tracking plays a key role in the information processing of spacebased optical systems. We propose a tracking method based on the multimodel cardinalized probability hypothesis density (CPHD) filtering. We first construct multiple motion models including a constant axial force model and a twobody gravity for describing the movement of LEO targets. We also build a spacebased measurement model. The extended state is established by combining the motion mode with the target state of the LEO target. Both the posterior probability hypothesis density (PHD) of the extended state and the posterior cardinality distribution of targets are propagated by using the CPHD filtering. Then, the target states and the target numbers are jointly estimated. Simulation results show that the proposed multimodel CPHD filtering achieves better estimation accuracy in target number and target states in comparison with the multimodel PHD filtering and the single model CPHD filtering.

Key words: spacebased optical observation;multimodel;CPHD filtering;LEO;multitarget tracking