北京邮电大学学报

  • EI核心期刊

北京邮电大学学报 ›› 2011, Vol. 34 ›› Issue (6): 125-128.doi: 10.13190/jbupt.201106.125.libb

• 研究报告 • 上一篇    

基于多假设高斯—厄密特滤波的红外单站目标跟踪算法

李彬彬1,冯新喜2,2,文曦2,2   

  1. 1. 空军工程大学电讯工程学院
    2.
  • 收稿日期:2010-12-10 修回日期:2011-04-22 出版日期:2011-12-28 发布日期:2011-10-18
  • 通讯作者: 李彬彬 E-mail:libinbin01132@163.com
  • 作者简介:李彬彬（1982-），女，博士生，E-mail：libinbin01132@163.com 冯新喜（1962-），男，教授，博士生导师

An Algorithm of IRST Single Observer Tracking Based on Multihypothesis Gauss-Hermite Filter

  • Received:2010-12-10 Revised:2011-04-22 Online:2011-12-28 Published:2011-10-18

PDF

947

摘要：

为了解决单红外观测站对目标进行跟踪时存在的非线性估计问题，提出了基于直角坐标系的多假设高斯-厄密特滤波算法。该算法为了减小距离的不可观测性对于非线性滤波的影响，在假设的多段距离间隔中采用并行的高斯-厄密特滤波并加权获得目标的状态估计。首先将初始时刻红外观测站的观测距离区间划分为若干个子区间，每个子区间表示关于目标真实距离的一种假设；在每个子区间内用共同的量测进行独立的高斯-厄密特滤波，并将每个子区间对应的概率根据贝叶斯规则进行递归计算；最后将各子区间的状态估计和协方差进行加权求和得到最终估计。仿真实验表明，由于高斯-厄密特滤波的使用避免了求解雅克比矩阵，并且距离的多假设降低了不确定性，该算法的滤波精度高于高斯-厄密特滤波和扩展卡尔曼滤波。

Abstract:

For the problem of nonlinear estimation with single IRST in target tracking, a new algorithm of Multihypothesis Cartesian Gauss-Hermite filter is presented. In order to decrease the influence of non-observability of distance, based on Gauss-Hermite filter, multihypothesis of distance is adopted to operate collateral filter. First, at the initial time, the measurement interval of IRST is divided into a number of subintervals, each of which represents a hypothesis responding to the true range of the object. For each hypothesis, Gauss-Hermite filter is implemented with the common measurement. The probability of each subinterval is calculated recursively according to Bayes’ criteria. Finally, the weighted estimation of state and covariance is calculated for sum respectively to get the final eastimation.. Simulation results show that the accuracy of filter in this paper is better than the original Gauss-Hermite filter and EKF ,for the reason that Gauss-Hermite filter avoids the calculation of Jacobian matrix and the multihypothesis of distance decreases uncertainty.

中图分类号: 

版权所有 © 2018 《北京邮电大学学报》编辑部 京ICP备14033833号     在线人数：今日访问 当前在线 总访问

通信地址：北京市海淀区西土城路10号北京邮电大学195信箱    邮编：100876    电话：(010)62281995    QQ：307514248    微博    留言    电子邮箱：byxb@bupt.edu.cn