摘要
针对非线性、非高斯系统的状态估计问题,本文提出了一种基于区间估计的粒子滤波算法.新算法从辅助粒子滤波算法的理论出发,首先对系统状态的期望值进行区间估计,然后在该区间上均匀采样,并利用当前观测信息进行修正,最后得出滤波结果.为了保证估计区间的有效性和算法计算效率,本文给出了区间扩展条件.由于算法直接在区间上均匀采样,不仅避免了重采样带来的样本贫化,而且保证了粒子的多样性.实验结果表明,该算法具有较高的滤波精度,明显优于一般的粒子滤波算法.
To deal with non-linear, non-Gaussian state estimation problem, a kind of particle filter algorithm based on interval estimation was proposed. This paper analyzed the auxiliary particle filter at first. After interval estimating the expectation of the system states, the new algorithm sampled uniformly in the interval and updated the filter results using the new measurement. The interval extension conditions were proposed to ensure the validity of the estimated range and computational efficiency of the algorithm. Sampling uniformly in the interval avoids the particle degeneracy and improves the particle divergence. The experimental results show that the new particle filter is significantly better than the general particle filter.
出处
《哈尔滨工业大学学报》
EI
CAS
CSCD
北大核心
2013年第11期8-12,共5页
Journal of Harbin Institute of Technology
基金
国家自然科学基金资助项目(61074127)
关键词
贝叶斯滤波
粒子滤波
区间估计
均匀采样
bayesian filtering
particle filter
interval estimate
uniform sampling
