粒子滤波算法

Particle Filtering Algorithm

粒子滤波算法逐渐成为科学领域的研究热点.文章首先阐述了粒子滤波算法的提出背景,根据m阶马尔科夫假设,分析算法基本原理并推导后验概率密度及权值更新公式.分析了基本粒子滤波算法中存在的问题以及解决方法.针对粒子滤波算法重要性采样密度的选择问题,综述了重要性采样密度选择方法.对重采样技术及样本匮乏问题进行了深入的分析,讨论了算法收敛性分析的最新进展.对自适应粒子滤波算法以及粒子滤波算法在各主要应用领域的进展进行了论述.最后对粒子滤波算法的研究前景提出了展望.

Particle filter is emerging as a new hotspot of research in scientific fields in the past several years.We first show the background information of particle filters.Thereafter,the principle of the particle filter under m-order Markovian assumption is analyzed,accompanying the derivatives of the posterior density function and the weight updating formula.Meanwhile,the analysis of the drawbacks of the standard particle filter and corresponding solutions are given.And a critical survey of importance sampling density selection is shown in the following section.We also give a detailed analysis of resampling method and the sample impoverishment problem induced by resampling.We reviewed the development of adaptive particle filters following the advances of convergence analysis.The following section reviews the advances of particle filters in different application areas.Finally,the future directions are pointed out.