Dynamic resampling particle filter adaptive to changes in system model  

作　　者：Ling WU Faxing LU 

机构地区：[1]Electronic Engineering College, Naval University of Engineering, Wuhan Hubei 430033, China

出　　处：《控制理论与应用（英文版）》2011年第4期584-588,共5页

基　　金：supported by the National Natural Science Foundation of Hubei Province (Nos. 2009CDB098, 2009CDB274)

摘　　要：A dynamic resampling particle filter （DRPF）, integrated with scatteriflg and migrating operations, is proposed for the flexibility in nonlinear systems subject to changes in the system model without a considerable increase in computational cost. Under the change detection with the tracking error, the scattering and migration are alternatively adopted where the scattering can boost the diversity in the particle population when no change, or slow change occurs, while the migrating updates particles with new observations if the change comes abruptly. Simulation results validate the proposed PF as a promising alternative to the existing PFs.A dynamic resampling particle filter （DRPF）, integrated with scatteriflg and migrating operations, is proposed for the flexibility in nonlinear systems subject to changes in the system model without a considerable increase in computational cost. Under the change detection with the tracking error, the scattering and migration are alternatively adopted where the scattering can boost the diversity in the particle population when no change, or slow change occurs, while the migrating updates particles with new observations if the change comes abruptly. Simulation results validate the proposed PF as a promising alternative to the existing PFs.

关 键 词：Particle filter Bayesian estimation Change detection RESAMPLING 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置] TP273.2[自动化与计算机技术—控制科学与工程]