文档介绍：系统仿真学报 Vol. 17 No. 10
2360 JOURNAL OF SYSTEM SIMULATION Oct. 2005
基于交互式多模型的粒子滤波算法
邓小龙谢剑英杨煜普
(上海交通大学自动化系, 上海 200030)
摘要: 融合交互式多模型和粒子滤波提出了一种新的多模型粒子滤波算法该算法采用多模型
结构以跟踪目标的任意机动各模型采用粒子滤波算法以处理非线性非高斯问题各模型中相
对固定数目的粒子群经过相互交互粒子滤波后再进行重抽样以减少滤波退化现象与通用的交互
式多模型算法进行了比较试验仿真结果证实了本文新滤波算法的有效性
关键词: 交互式多模型粒子滤波非线性非高斯重抽样
文章编号 1004-731X (2005) 10-2360-03 中图分类号 TP273 文献标识码 A
Particle Filter Based on Interacting Multiple Model
DENG Xiao-long, XIE Jian-ying, YANG Yu-pu
(Department of Automation of Shanghai Jiaotong University, Shanghai 200030, China)
Abstract: Combining interacting multiple model with particle filter, a new multiple model particle filter is presented. The
algorithm used the multiple models to track arbitrary maneuvering of the target. Every model used particle filter to deal with
the nonlinear and non-Gaussian problems. After interaction and particle filtering, particles in the models with the fixed
number are resampled to reduce the degeneracy of filtering. In the simulations, compared with the general interacting
multiple model, the results demonstrate the efficiency of the new filtering method.
Key words: interacting multiple model; particle filter; nonlinear / non-Gaussian; resampling
本文提出了一种新的融合和粒子滤波的滤波方
引言1 IMM
法新方法采用多模型结构各模型的滤波方法采用粒子滤
在机动目标跟踪领域次优的基于跳跃马尔可夫线性系波算法各模型中的粒子数目固定且独立于模型概率以充分
统的多模型滤波算法得到了广泛的关注如广义伪贝叶斯算体现多模型的特点模型间的粒