摘要
在整个导弹防御系统中,多目标跟踪是很重要的一项技术,要求系统快速机动地跟踪导弹目标,但系统存在非线性问题,使用传统方法使跟踪偏差大。为解决上述问题,提出在非高斯条件下,把高斯-厄米特粒子滤波算法和联合概率数据关联方法相结合,对多目标跟踪的数据进行关联处理并进行状态估计。利用高斯-厄米特滤波计算的均值、协方差产生密度函数,并生成具有后验特征的粒子。用联合概率数据关联方法进行杂波剔除和数据关联,并对综合的关联粒子滤波算法进行仿真。仿真结果表明,改进方法可以有效解决多目标的准确跟踪问题。
In missile defense system, multiple targets tracking is a key technology. For multiple missile targets of fast maneuvering, this paper integrated the Gauss-Hermite particle filter and Joint probabilistic data association method together, and then processed multi-target tracking data associationly and estimateed state value in the condi-tion of nonlinear and non-Gaussian. This method used Gauss-Hermite filter which calculates the mean and covari-ance to generate the importance density function, and then generated particles with posterior characteristics. Mean-while, joint probabilistic data association method was used to eliminate clutter waves and associate data. Finally, as-sociation particle filter algorithms was generated. Simulation results show that the proposed method can solve the prob-lem of tracking multiple targets effectively.
出处
《计算机仿真》
CSCD
北大核心
2012年第9期17-21,共5页
Computer Simulation
关键词
多目标跟踪
高斯-厄米特积分
联合概率数据关联
重要密度函数
关联粒子滤波
Multiple targets tracking
Gauss-hermite particle filter
Joint probabilistic data association
Importancedensity function
Association particle filter