摘要
主要对粒子滤波算法进行综述。首先详细描述了递归贝叶斯估计的基本原理和基于蒙特卡罗方法的重要性采样/重采样技术,在此基础上引出了粒子滤波标准算法——序贯重要性采样算法和序贯重要性重采样算法。针对这两个算法在应用中存在的问题,从提高算法的有效性和实时性两个方面,对近年来国内外在粒子滤波理论及应用研究方面开展的工作进行了介绍、分析归纳了改进粒子滤波算法及其主要改进思想。最后,对粒子滤波算法的研究方向进行了展望。
This paper reviews the algorithms for particle filtering. We first present the basic knowledge of the Recursive Bayesian Estimation, and the Importance Sampling/Resampling Techniques which are based on the Monte Carlo Method. We then describe in detail the two standard algorithms, Sequential Importance Sampling and Sequential Importance Resampling. We subsequently introduce recent progress in the research of particle-filtering algorithms to tackle issues encountered by the two standard algorithms in practical applications. We analyze key ideas of the new algorithms. We classify the new algorithms according to their improvements in the validity and real-time performance. We finally give some suggestions of future research directions.
出处
《天文研究与技术》
CSCD
2013年第4期397-409,共13页
Astronomical Research & Technology
关键词
数据处理
粒子滤波算法
综述
非线性滤波
递归贝叶斯估计
重要性采样
重采样
Data processing
Particle-filtering algorithm
Review
Nonlinear filter
Recursive Bayesian Estimation
Importance Sampling/Resampling
