摘要
针对标准粒子滤波算法存在的粒子退化与贫化问题,提出了一种新的改进粒子滤波算法。该算法采用无迹卡尔曼滤波、优化组合策略和标准粒子滤波相结合的方法,运用UKF产生重要性密度函数,解决标准PF算法中以先验概率密度函数作为建议分布所引发的退化问题;运用优化组合重采样策略保证所有粒子的信息以一定概率得到继承,维持粒子集中粒子的多样性。理论分析与仿真结果均表明,改进算法能有效地解决标准粒子滤波存在的粒子退化问题并避免粒子贫化现象的出现,具有更高的状态估计精度。
In order to solve particle degeneracy and simultaneously avoid sample impoverishment, we propose a new improved particle filter algorithm based on the unscented Kalman filter （UKF）, optimized combination strategy, and the standard particle filter method. We use the UKF to generate the importance density function and solve all the problems caused by the traditional particle filters which use prior density function as the particle distribution. And then we employ the optimized combination scheme to ensure all useful information inherited, which can maintain particle diversity. Theoretical analysis and simulation results both show that the improved particle filter algorithm can solve particle degeneracy and avoid sample impoverishment, and it has higher filtering accuracy in state estimation.
出处
《计算机工程与科学》
CSCD
北大核心
2017年第8期1483-1488,共6页
Computer Engineering & Science
关键词
粒子滤波
无迹卡尔曼滤波
优化组合策略
距离判决
particle filter
unscented Kalman filter
optimized combination scheme
distance comparing
