基于分层转移的粒子滤波MCMC重采样算法
DOI:
CSTR:
作者:
作者单位:

徐州工程学院信电学院

作者简介:

田隽

通讯作者:

中图分类号:

基金项目:

国家高技术研究发展计划(863)项目;江苏省产学研联合创新基金项目


Resampling algorithm for particle filter based on layered transacting MCMC
Author:
Affiliation:

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    针对粒子滤波中如何设计重采样策略以解决“权值蜕化”, 同时又可避免“样本贫化” 的问题, 提出一
    种基于分层转移的Monte Carlo Markov 链(MCMC) 重采样算法. 当样本容量检测出现“蜕化” 时, 将样本集按权值
    蜕化程度进行分层, 利用提出的变异繁殖算法, 将其与PSO 融合产生MCMC转移核, 并施以分层子集; 然后通过
    Metroplis-Hastings 算法进行接收-拒绝采样, 由此构建的Markov 链可收敛到与目标真实后验等价的平稳分布. 数值
    仿真结果表明, 所提出的算法能以更快的收敛速度和更小的估计误差贴近目标真实后验, 从而提高了估计精度.

    Abstract:

    To resolve weight degeneracy and avoid sample impoverishment in resampling algorithms of particle filter,a
    method, named layered transacting MCMC-resampling algorithm, is proposed. When the effective sample size is below a
    fixed threshold, particles are dived into two sample subsets according to their individual weights. Mutation operator and
    PSO, which are considered as transition kernels of MCMC, are applied to sample subsets respectively. Then an acceptancerejection
    rule of Metropolis-Hastings algorithm is used to generate the Markov chain with the stationary distribution which
    is equivalent to target posterior density. The simulation results show that the proposed method is superior to other resampling
    algorithms both in accuracy and convergence speed.

    参考文献
    相似文献
    引证文献
引用本文

田隽, 钱建生, 李世银.基于分层转移的粒子滤波MCMC重采样算法[J].控制与决策,2011,26(8):1253-1258

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2010-06-01
  • 最后修改日期:2010-09-17
  • 录用日期:
  • 在线发布日期: 2011-08-20
  • 出版日期:
文章二维码