改进渐消中心差分粒子滤波的单站无源定位算法  

Adaptive fading central difference particle filter algorithm for single observer passive location

作　　者：宋海亮[1,2] 付永庆[1] 刘学[3] 

机构地区：[1]哈尔滨工程大学信息与通信工程学院 [2]中国人民解放军91868部队 [3]中国人民解放军91550部队94分队

出　　处：《应用基础与工程科学学报》2015年第4期842-849,共8页Journal of Basic Science and Engineering

基　　金：国家自然科学基金项目资助(61172038,60772025)

摘　　要：针对粒子滤波中存在的重要性密度函数难以选取和可能出现粒子退化的问题,提出了一种改进渐消中心差分粒子滤波算法.该算法从研究粒子滤波的建议分布出发,充分利用最新观测信息,通过引入改进的渐消因子实时调整增益阵,得到一种优化的建议分布函数,能够有效缓解粒子退化现象.同时,在粒子采样中,研究采用了多样化的采样方式减轻滤波中粒子枯竭的影响,进一步提高了滤波精度.将该算法应用到单站无源定位系统中进行仿真运算,仿真结果表明,在不同的观测精度环境下,该算法使得系统具有更好的自适应性和滤波精度.A novel adaptive fading central difference particle filter algorithm（ AFCDPF） is presented to solve the problems of selection about important density function and particle degeneracy. This algorithm uses modified central difference kalman filter（ CDKF） to produce proposal distribution function to weaken the impact of degeneration,and the adaptive factor is introduced to improve filtering precision and overcome the limitation of the general particle filtering by fully using observation information. The diversity of samples is adopted to mitigate the effect of particle impoverishment and improve the filtering precision. The proposed algorithm is applied to the single observer passive location system（ SOPLS）. The simulation results show that the improved algorithm has faster convergence,higher stability and accuracy.

关 键 词：改进渐消 中心差分卡尔曼滤波 粒子滤波 多样化采样 单站无源定位 

分 类 号：N95[自然科学总论]