摘要
针对粒子滤波算法中存在的粒子退化与粒子匮乏的缺陷,提出了利用高斯粒子群优化无迹粒子滤波的新算法。算法使用无迹粒子滤波进行重要性采样,并将高斯粒子群优化算法融入重采样过程中。该算法选取的概率密度更加接近系统真实状态,有效增加了粒子的多样性,提高了抽样效率,降低了粒子退化程度,缓解了粒子匮乏现象。试验结果表明,该算法的滤波精度明显优于粒子滤波与无迹粒子滤波算法所得到的滤波精度。
A new unscented particle filter using Gaussian particle swarm optimization (GPSO-UPF) algorithm is proposed in this paper to improve particle degeneracy and particle impoverishment. It uses unscented particle filter in importance sampling process and incorporates Gaussian particle swarm optimization into re-sampling process. Through GPSO-UPF, the probability density moves closely to true state, the number of effective particles and efficiency are increased, the particle degeneracy is reduced and particle impoverishment is relieved. The experimental results show that the state estimation precision of GPSO-UPF is higher than estimation precision of PF and UPF.
出处
《测绘通报》
CSCD
北大核心
2017年第4期1-5,共5页
Bulletin of Surveying and Mapping
基金
国家重点基础研究发展计划(2013CB733205)
武汉市科技局项目(2015011701011639)
关键词
无迹粒子滤波
高斯粒子群优化
粒子退化
粒子匮乏
unscented particle filter
Gaussian particle swarm optimization
particle degeneracy
particle impoverishment