混沌映射采样的粒子滤波器  

Chaos map sampling particle filtering

作　　者：宁小磊[1] 陈战旗 赵新[1] 何星[2] 石国祥[2] 

机构地区：[1]中国华阴兵器试验中心,陕西华阴714200 [2]第二炮兵工程学院304教研室,西安710025

出　　处：《控制与决策》2011年第12期1777-1782,共6页Control and Decision

基　　金：装备预研基金项目(2009SY3213001;51309060302)

摘　　要：针对粒子滤波粒子多样性减弱引起的粒子枯竭问题,提出一种新的基于混沌映射采样的粒子滤波改进算法(CMS-PF).在重要性采样之后,用类似载波的方法将离线生成的混沌序列映射到以较大权重粒子为中心的样本子空间,从而生成一些映射粒子,并结合当前时刻的预测粒子构建候选粒子集,最终依据各粒子自身的权重实现优选.仿真结果表明,该算法能有效提高对非线性系统状态的估计精度.A novel particle filter named chaos map sampling particle filter（CMS-PF） is proposed to solve the particle impoverishment problem. After important sampling, chaos series obtained offline are transformed into some sub-spaces, whose kernels are the particles with heavy weights, by using the algorithm similar to carder wave to generate map particles. The map particles are associated with predictive particles to construct the candidate particle set and the optimizing selection of particles is realized based on its own weighs. Simulation results show that this method can effectively improve the state estimation precision.

关 键 词：粒子滤波 混沌映射 粒子枯竭 非线性估计 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]