多移动观测平台纯角度目标跟踪融合算法研究

发布时间：2018-04-27 04:10

本文选题：目标跟踪 + 纯角度　；参考：《南京理工大学》2017年硕士论文

【摘要】：纯角度目标跟踪,由于观测平台无需发射电磁波等信号,只需被动接收目标客观存在的辐射信息或者是目标对其它光源的反射能量,就可以定位并跟踪目标,因此成为国内外学者研究的热门课题。纯角度跟踪系统具备很强的隐蔽性和生存能力,而多移动观测平台纯角度目标跟踪系统更是具有搜索范围大、作用距离远和可靠性高等显著特点,特别适合于广阔海域和空域的监视。本论文对多移动观测平台纯角度目标跟踪融合算法进行了比较深入的研究,提出了一些新的思路和方法,所做工作及成果如下:(1)研究了三维空间中多观测平台纯角度目标跟踪的系统框架,建立了几种典型的目标运动模型和量测模型。基于所建立的系统模型,讨论了 Monte Carlo法,定义了均方根误差(RMSE)等常用的目标跟踪性能评估指标。(2)推导了三维空间多观测平台测向交叉定位最小二乘算法公式系,引入几何精度因子(GDOP)刻画定位精度,并给出了三维空间多观测平台目标定位的GDOP解析表达式。分别针对非机动目标和机动目标进行了交叉定位跟踪的仿真验证分析,绘制了GDOP分布图,结果表明,所提出的算法可以实现对目标的有效跟踪。(3)对比分析了扩展卡尔曼滤波(EKF)和无迹卡尔曼滤波(UKF)两种滤波算法的性能,提出了将平方根无迹卡尔曼滤波算法(SR-UKF)用于多观测平台纯角度融合跟踪系统。基于SR-UKF,提出了一种先量测融合、后航迹融合的分布式融合跟踪算法,相对EKF和UKF,该算法能适应跟踪精度和稳定性要求更高的目标跟踪系统,实验结果验证了该算法的有效性。(4)针对传感器网络观测系统,提出了一种基于最小GDOP的观测平台优化选择方法。该方法首先根据交叉定位得到的目标粗略位置,通过最小化GDOP,利用粒子群优化(PSO)算法快速搜索最优传感器组合,结合SR-UKF滤波算法解算出更精确的目标位置,并根据目标飞行轨迹的变化动态调整参与观测的传感器。理论分析和仿真实验验证了该方法的可行性和有效性,对大规模传感器网络目标跟踪系统中的传感器优化调度具有较强的适用性。(5)针对观测平台可移动的情况,提出了一种基于最小GDOP的观测平台机动策略。针对静止和运动目标,分别给出了可移动观测平台的机动策略。仿真结果表明,本论文提出的观测平台机动策略提高了目标跟踪系统的鲁棒性和稳定性。在海对空、空对空和路径约束下的地对空三种目标跟踪场景下,均能实现有效跟踪,且目标跟踪性能优于静止观测平台跟踪系统。
[Abstract]:The pure angle target tracking, because the observation platform does not need to transmit the electromagnetic wave and so on signal, only needs passively receives the target objective existence radiation information or the target to other light source reflection energy, can locate and track the target, Therefore, it has become a hot topic for scholars at home and abroad. The pure angle tracking system has strong concealment and survivability, while the pure angle target tracking system of multiple mobile observation platforms has the remarkable characteristics of large search range, long range of action and high reliability, etc. It is particularly suitable for surveillance of vast sea areas and airspace. In this paper, the fusion algorithm of pure angle target tracking for multi-mobile observation platform is deeply studied, and some new ideas and methods are put forward. The work and results are as follows: (1) the system framework of pure angle target tracking for multi-observation platform in three-dimensional space is studied, and several typical target motion models and measurement models are established. Based on the established system model, the Monte Carlo method is discussed, and the commonly used target tracking performance evaluation index (RMSE), such as root mean square error (RMSE), is defined. The geometric precision factor (GDOP) is introduced to depict the positioning accuracy, and the GDOP analytical expression for the target location of the three-dimensional multi-observation platform is given. The simulation verification and analysis of cross positioning and tracking for non-maneuvering targets and maneuvering targets are carried out respectively, and the GDOP distribution diagram is drawn. The results show that, The proposed algorithm can realize the effective tracking of the target. The performance of the extended Kalman filter (EKF) and the unscented Kalman filter (UKF) are compared and analyzed. A square root unscented Kalman filter algorithm named SR-UKF is proposed for the pure angle fusion tracking system of multi-observation platform. Based on SR-UKF, a distributed fusion tracking algorithm based on measurement fusion and track fusion is proposed. Compared with EKF and UKF, this algorithm can adapt to target tracking system with higher tracking accuracy and stability. Experimental results show that the algorithm is effective. (4) an optimal selection method based on minimum GDOP is proposed for sensor network observation system. Firstly, according to the rough position of the target obtained by cross-location, the particle swarm optimization (PSO) algorithm is used to quickly search the optimal sensor combination by minimizing the GDOP, and the SR-UKF filtering algorithm is used to calculate the more accurate target position. The sensors involved in the observation are dynamically adjusted according to the change of the target's flight trajectory. The theoretical analysis and simulation results show that the method is feasible and effective. It is applicable to the optimal scheduling of sensors in large-scale sensor network target tracking systems. An observational platform maneuver strategy based on minimum GDOP is proposed. For stationary and moving targets, the maneuvering strategies of the mobile observation platform are given respectively. Simulation results show that the maneuvering strategy proposed in this paper improves the robustness and stability of the target tracking system. In the ground to air target tracking scenarios with sea to air, air to air and path constraints, the tracking performance of the target is better than that of the static observation platform tracking system.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP212.9;TN713

【相似文献】