粒子滤波进展与展望  

Development and prospect of particle filtering

作　　者：杨小军[1] 潘泉[1] 王睿[1] 张洪才[1] 

机构地区：[1]西北工业大学自动化学院

出　　处：《控制理论与应用》2006年第2期261-267,共7页Control Theory & Applications

基　　金：国家自然科学基金资助项目(60404011;60372085)

摘　　要：粒子滤波器是基于序贯M onte Carlo仿真方法的非线性滤波算法,本文对粒子滤波器的研究现状和研究进展做了综述,详细论述了粒子滤波原理、收敛性、应用及进展.首先在Bayes框架内分析了序贯重要性采样原理,重要性分布函数的选择,以及重采样方法,总结了粒子滤波器发展过程中的各种改进策略和新变种,讨论了粒子滤波器在各个领域的应用及进展,最后介绍了粒子方法的新发展,新动态,并对未来发展方向做了进一步的展望.Particle filtering is a sequential Monte Carlo simulation based on nonlinear filtering algorithm. An overview of the status and development of research on particle filtering is presented. The principle, convergence, application and evolution of particle filtering are described in detail. First, the principle of sequential importance-sampling, the choice of importance distribution function, and the method of re-sampling are analyzed within Bayesian framework. Secondly, the improvement methods and novel variations of particle filtering are then summarized. Thirdly, the application and development in various areas are reviewed. Fourthly, the novel extension and trends of particle filtering are illustrated. Finally, further research prospects are introduced.

关 键 词：BAYES估计 粒子滤波器 最优滤波 序贯Monte CARLO方法 

分 类 号：TN713[电子电信—电路与系统]