基于高斯混合滤波的目标跟踪方法

发布时间：2018-01-27 09:46

本文关键词： 目标跟踪模糊多普勒量测无迹卡尔曼滤波高斯和群粒子　出处：《哈尔滨工业大学》2014年硕士论文　论文类型：学位论文

【摘要】：雷达目标跟踪[1]中，一般目标的状态建立在笛卡尔坐标系中，而目标的量测信息是在球坐标或极坐标下给出。球坐标系下的无噪声无模糊多普勒量测信息在笛卡尔坐标系下使用是一种非线性的雷达目标跟踪问题。为了解决这个问题本文首先叙述了最常用的目标跟踪方法[2]，比如卡尔曼滤波[3]、无迹卡尔曼滤波[4]。其次，，在无迹卡尔曼滤波的基础上文章提出了高斯和无迹卡尔曼滤波。用高斯和表示模糊多普勒量测的概率密度函数。基于高斯和的无迹卡尔曼滤波方法解决模糊的多普勒量测状态估计问题。仿真结果说明了这种方法的有效性。此方法与无迹卡尔曼滤波器以及转换量测卡尔曼滤波器比较显示在位置和速度误差方面有更高的精度。最后，着重讲解了一下高斯和参数的估计问题。因为高斯和是上面我们提到的滤波算法的基础。本论文也提出了两种最有效的方法EM算法和群粒子（PSO）算法。其中EM算法也称为期望最大化算法，主要通过期望步骤和期望最大化步骤不断地更新得到的估计值，直到达到一个我们认为合理的结果。粒子群方法估计高斯和参数主要是用群搜索的特性完成的。在特定的范围内通过个体和群体之间其他个体的相互协作完成最优估计的搜索。
[Abstract]:Radar target tracking. [In [1], the state of the general object is established in the Cartesian coordinate system. The measurement information of the target is given in the spherical coordinate or polar coordinate. It is a nonlinear radar target tracking problem that the noiseless non-fuzzy Doppler measurement information is used in the Cartesian coordinate system. In order to solve this problem, this paper first describes the most commonly used target tracking methods. [2], for example, Kalman filtering. [3], unscented Kalman filter. [4]. Second. On the basis of unscented Kalman filter, Gao Si and unscented Kalman filter are proposed. The probability density function of fuzzy Doppler measurement is represented by the Gao Si sum. The unscented Kalman filter method based on Gao Si sum is used to solve fuzzy multiple. The simulation results show that this method is effective. Compared with the unscented Kalman filter and the converted measurement Kalman filter, the proposed method has higher accuracy in position and velocity error. Degree. Finally. This paper focuses on the estimation of Gao Si and its parameters. Because Gao Si sum is the basis of the filtering algorithm mentioned above, this paper also proposes two most effective methods, EM algorithm and Swarm PSO). The EM algorithm is also called the expectation maximization algorithm. The resulting estimates are updated mainly through the expected steps and the expectation maximization steps. Until we reach a reasonable result. Particle swarm optimization method estimates Gao Si and parameters are mainly done with the characteristics of swarm search. In a specific range of individuals and other individuals between the cooperation of the completion of the most. Search for optimal estimates.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN953

【相似文献】