基于被动多传感器的目标跟踪方法研究

项目名称： 基于被动多传感器的目标跟踪方法研究

项目编号： No.60871074

项目类型： 面上项目

立项/批准年度： 2009

项目学科： 无线电电子学、电信技术

项目作者： 姬红兵

作者单位： 西安电子科技大学

项目金额： 35万元

中文摘要： 被动传感器探测因其良好的隐蔽性、抗攻击和抗干扰等优点，已成为现代防御系统的研究热点。本项目重点研究被动多传感器下的目标跟踪及量测数据关联等关键问题，取得了一系列创新性研究成果。（1）针对单目标跟踪，研究了高斯滤波和粒子滤波，提出的修正IEKF以及QMC-GPF算法有效改善了观测非线性条件下的滤波性能。（2）针对机动目标跟踪，提出了基于交互多模型、曲线模型自适应以及修正输入估计等算法，有效提高了强机动目标跟踪精度。（3）针对多目标跟踪，重点研究了基于随机集理论的多目标跟踪算法。改进的多模型粒子PHD算法实现了PHD框架下的输入交互，改进的CPHD滤波算法通过对权值进行动态分配有效解决了传统CPHD中固有的目标漏检问题，基于模糊聚类的航迹维持算法通过引入多帧信息有效提高了随机集滤波的航迹维持性能。（4）针对量测数据关联，改进的拉格朗日松弛算法通过修正代价函数以及优化松弛过程，有效提高了关联精度和关联效率。本项目圆满完成了项目申请书既定的研究内容，研究成果对于构建被动多传感器目标跟踪的理论体系具有重要的意义，并在武器装备和防御系统的推广应用方面具有重要的价值。

中文关键词： 被动多传感器；目标跟踪；粒子滤波；随机集

英文摘要： Because of the merit of concealment and capability to avoid attack, passive detecting and tracking have been one of the hot topics in the modern defense systems. Our research is focus on the target tracking and passive sensor data association problems in the passive multi-sensor systems, and many contributions have been made. (1) Aiming at the single target tracking problem, some new nonlinear filters based on Gaussian filters or particle filters have been proposed, and the modified IEKF and the QMC-GPF can solve the nonlinear problem and improve the tracking accuracy . (2) Aiming at the maneuvering target tracking problem, some new algorithms based on interacting multiple models, adaptive filters, and modified input estimations have been proposed to improve the maneuvering target tracking performance. (3) Aiming at the multi-target tracking problem, some new algorithms have been proposed with emphases on the methods based on the random finite set theory. The improved multiple model PHD filter extends the multiple model to the interacting multiple model in the framework of PHD filter. The improved CPHD filter solves the missed detection problem in the conventional CPHD filter by employing a dynamic reweighting scheme. The fuzzy clustering based algorithm improves the performance of track continuity by taking the entire multi-frame information into account. (4) Aiming at the passive sensor data association problem, the improved Lagrangian relaxation algorithm improves the correlation accuracy and the computational efficiency by modifying the cost function and optimizing the relaxation process, respectively. To summarize, all the tasks in the fund application have been completed, and our research contributions will have significant effect in the passive multi-sensor target tracking theory, and have potential value to improve the performances of weapons and equipments in the modern defense systems.

英文关键词： Passive multi-sensor; Target tracking; Particle filter; Random finite set