一种改进残差重采样算法的研究  

Improving the residual resampling algorithm

作　　者：冯驰[1] 赵娜[1] 王萌[1] 

机构地区：[1]哈尔滨工程大学信息与通信工程学院,黑龙江哈尔滨150001

出　　处：《哈尔滨工程大学学报》2010年第1期120-124,共5页Journal of Harbin Engineering University

基　　金：国家自然科学基金资助项目(60873037);黑龙江省教育厅科技基金资助项目(11511118)

摘　　要：粒子滤波算法对非线性非高斯系统有很好的估计性能,但是存在较为严重的退化问题.重采样算法的提出,有效缓解了粒子滤波器的退化问题.但重采样算法本身也存在一定问题,针对残差重采样和残差系统重采样算法进行研究,并提出改进算法.该改进算法避免了残差重采样算法中的残留粒子重采样问题,减少了运算量,提高了运行效率.仿真结果表明该算法的运行效率明显高于残差重采样和残差系统重采样2种算法,并且随着粒子数目的增加,这种优势表现地更加显著.Particle filters perform well when estimating nonlinear/non-Gaussian systems,yet serious degeneration can occur.Resampling algorithms have been used to alleviate particle degeneracy.Based on studies of residual resampling and residual systematic resampling,the authors proposed improved methods for residual resampling.This improved algorithm avoids resampling residual particles in the residual resampling,thus reducing computational complexity and improving running efficiency.Simulation results showed that the running efficiency of this improved algorithm was obviously higher than that of residual resampling and residual systematic resampling.When there were many particles,its efficiency was outstanding.

关 键 词：粒子滤波器 重采样 改进残差重采样 运行效率 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]