非线性量测下的雷达目标跟踪算法研究

发布时间：2018-06-29 12:38

本文选题：非线性滤波 + 目标跟踪　；参考：《电子科技大学》2017年硕士论文

【摘要】：目标跟踪已经被广泛的运用到军用和民用领域,滤波算法在各种目标跟踪系统里面都扮演着非常重要的角色。但是在大多数建立在极坐标或球坐标下的雷达实时跟踪系统中,系统的量测方程是非线性的。这超出了标准卡尔曼滤波算法的处理范围,在这种情况下,非线性滤波算法就被派上了用场。针对这个既实际又有必要深入研究的问题,本文对雷达目标跟踪中使用的多种基于非线性量测信息的滤波算法展开了研究。首先,本文研究了扩展卡尔曼滤波算法、无迹卡尔曼滤波算法以及粒子滤波算法这几种典型的非线性滤波算法。并且使用仿真实验从目标跟踪精度以及算法运算复杂度方面来比较了这三种算法。其次,在仅利用雷达系统获得的目标位置量测信息的条件下,本文研究了基于量测转换的卡尔曼滤波算法,包括无偏量测转换卡尔曼滤波算法、修改的无偏量测转换卡尔曼滤波算法以及去相关无偏量测转换卡尔曼滤波算法。本章同样也利用仿真实验从目标跟踪精度以及算法运算复杂度方面来比较了这三种算法。然后考虑系统能够同时获得目标的多普勒信息的目标跟踪场景。本章首先介绍了两种能利用多普勒信息的目标跟踪算法,包括基于量测值的序贯量测转换卡尔曼滤波算法,和近期提出的静态融合多普勒量测转换卡尔曼滤波算法。本文将DUCMKF更进一步,提出了基于预测值的序贯量测转换卡尔曼滤波算法。同样,本章也通过利用仿真实验来从目标跟踪精度以及算法运算复杂度方面来比较了这三种算法,表明新算法能获得更高的目标跟踪精度,其复杂度仅略微增加。最后研究了基于BLUE的雷达目标跟踪算法。首先简要介绍了BLUE,然后通过对BLUE算法步骤的一系列等价变换,推导出了卡尔曼框架结构形式下的BLUE算法。接着根据BLUE算法的卡尔曼滤波形式和卡尔曼算法的相似性,将BLUE算法和SQ-DUCMKF结合起来,提出了序贯的多普勒量测BLUE算法,将BLUE算法扩展到了能利用多普勒信息的场景中。仿真结果表明,BLUE和DUCMKF的性能相当,SQ-BLUE和SQ-DUCMKF的性能相当,能实现在位置量测信息下以及同时具有多普勒量测信息时的高精度目标跟踪。
[Abstract]:Target tracking has been widely used in military and civil fields. Filtering algorithm plays a very important role in various target tracking systems. However, in most radar real-time tracking systems based on polar or spherical coordinates, the measurement equations are nonlinear. This is beyond the scope of the standard Kalman filtering algorithm, in which case the nonlinear filtering algorithm is used. Aiming at this problem, which is both practical and necessary to be studied deeply, this paper studies many filtering algorithms based on nonlinear measurement information used in radar target tracking. Firstly, several typical nonlinear filtering algorithms, such as extended Kalman filter, unscented Kalman filter and particle filter, are studied. Simulation experiments are used to compare these three algorithms in terms of target tracking accuracy and computational complexity. Secondly, under the condition of only using the target position measurement information obtained by radar system, this paper studies the Kalman filtering algorithm based on measurement conversion, including unbiased measurement conversion Kalman filter algorithm. Modified unbiased measurement conversion Kalman filter algorithm and uncorrelated unbiased measurement conversion Kalman filter algorithm. In this chapter, simulation experiments are also used to compare these three algorithms in terms of target tracking accuracy and computational complexity. Then consider the target tracking scene where the system can simultaneously obtain the Doppler information of the target. In this chapter, we first introduce two kinds of target tracking algorithms which can use Doppler information, including sequential measurement conversion Kalman filtering algorithm based on measurement value, and static fusion Doppler measurement conversion Kalman filter algorithm proposed recently. In this paper, DUCMKF is further studied, and a sequential measurement conversion Kalman filter algorithm based on predictive values is proposed. In the same way, the simulation experiments are used to compare these three algorithms in terms of target tracking accuracy and computational complexity. It shows that the new algorithm can achieve higher target tracking accuracy and its complexity is only slightly increased. Finally, the radar target tracking algorithm based on blue is studied. First of all, this paper briefly introduces Blue, and then deduces the blue algorithm in the form of Kalman frame structure by a series of equivalent transformations of blue algorithm steps. Then according to the similarity between the Kalman filter form of blue algorithm and the Kalman algorithm, combining blue algorithm with SQ-DUCMKF algorithm, a sequential Doppler measurement blue algorithm is proposed, and the blue algorithm is extended to the scene where Doppler information can be used. The simulation results show that the performance of blue and DUCMKF is comparable to that of SQ-BLUE and SQ-DUCMKF, and can achieve high precision target tracking under position measurement information and with Doppler measurement information at the same time.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN953;TN713

【参考文献】