基于ERTS平滑建议分布函数的粒子滤波新方法  

PARTICLE FILTERING METHOD BASED ON PROPOSAL DISTRIBUTION FUNCTION OF ERTS SMOOTHING

作　　者：李保利[1,2] 宫轶松[1] 张锦军[3] 李启高[4] 

机构地区：[1]解放军信息工程大学测绘学院 [2]总参测绘局 [3]92493部队22分队 [4]黄河水利委员会三门峡库区水文水资源局

出　　处：《大地测量与地球动力学》2010年第2期152-155,共4页Journal of Geodesy and Geodynamics

基　　金：国家自然科学基金(40974009;40474007)

摘　　要：提出一种新的建议分布函数的设计方案,即将扩展Kalman滤波算法与Rauch-Tung-Streibel(RTS)最优固定区间平滑算法融合,产生建议分布函数。与用扩展Kalman滤波产生建议分布函数的粒子滤波方法相比,该方法可更准确地给出状态的估计结果,从而使得系统具有更好的稳健性。通过对GPS/DR组合导航系统的试验,验证了该方法的有效性。As it is quite important to choose a reasonable proposal distribution function for the particle fihering,a new proposal distribution function is designed. This scheme fuses the extended Kalman filter and the RTS fixed interval smoothing to generate the proposal distribution function. Compared to the particle filtering with only the proposal distribution function of extended Kalman filter, this method can give more accurate estimation results, and make the system more robust. An experiment for the GPS/DR integrated simulation system validates the effectiveness of the new approach.

关 键 词：非线性估计 粒子滤波 扩展Kalman滤波 RTS固定区间平滑 GPS/DR组合导航 

分 类 号：P207[天文地球—测绘科学与技术]