摘要
粒子滤波(PF)算法存在的主要问题是粒子退化现象,利用重抽样过程可以有效减轻退化现象,但带来了采样枯竭问题,导致滤波精度下降.本文提出一种多样性引导的进化粒子滤波(DEPF),把粒子群优化(PSO)算法引入到传统PF中,通过PSO搜索寻优重新分配粒子,使粒子的表示更加接近真实后验,并在PSO的搜索寻优过程中使用多样性引导机制来保证所得粒子集的多样性,以提高PF的精度.仿真实验结果表明了该算法的有效性.
Degeneracy phenomenon is a main problem in particle filter (PF). Although the resampling process can be used to reduce the effects of degeneracy phenomenon,it produces the sample impoverishment problem which makes filter's performance worse. A diversity-guided evolutionary particle filter (DEPF) is proposed in this paper. To improve the performance of PF,particle swarm optimization (PSO) algorithm is introduced to form new particle filter ,in which the search and optimization ability of PSO are used to redistribute particles closer to the true posterior. Moreover,in the process of PSO,a diversity-guided mechanism is used to guarantee the diversity of particle set. The simulation results show the effectiveness of DEPF.
出处
《小型微型计算机系统》
CSCD
北大核心
2008年第5期867-870,共4页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(60575023)资助
安徽省自然科学基金项目(070412064)资助
合肥工业大学校科学研究发展基金项目(070504F)资助
关键词
粒子滤波
采样枯竭
粒子群优化算法
多样性
particle filter
sample impoverishment
particle swarm optimization
diversity