高斯辅助粒子算法  

作　　者：宋昊霖 

机构地区：[1]沈阳建筑大学,辽宁沈阳110168

出　　处：《科技资讯》2019年第1期73-75,共3页Science & Technology Information

基　　金：辽宁省高等学校基本科研项目(项目编号:LJZ2017020);辽宁省自然科学基金面上项目(项目编号:20180520037);沈阳建筑大学一般项目(项目编号:2017028)

摘　　要：粒子滤波算法在处理非线性,非高斯问题的状态估计方面有独特的优势。粒子滤波方法由于存在粒子退化,粒子多样性匮乏等问题,不能准确地估计系统状态。该文提出了一种新的粒子滤波算法-高斯辅助粒子滤波算法,将即时的观测值信息引入重要性概率密度函数当中。该研究进行重采样时,用高斯变换对粒子进行了重新分布并调整重采样后的粒子权值。通过仿真实验验证该方法可以有效运用当前观测量值,优化粒子分布,在解决粒子退化问题的同时,也能维持粒子的多样性,提高滤波效率。Particle filter algorithm has a unique advantage in dealing with nonlinear,non-Gaussian state estimation.Particle filter method cannot accurately estimate the state of the system due to the particle degeneration and lack of particle diversity.In this paper,we propose a new particle f ilter algorithm called"Gauss based auxiliary particle filter"that introduces the real-time observation information into the importance probability density function.During resampling step,the particles are redistributed using Gaussian transformation and the weights of the particles after resampling are adjusted.The simulation results illustrate that the proposed method can effectively use the current values and optimize the particle distribution.This algorithm cannot only solve the problem of particle degeneration,but it can also maintain the diversity of particles and improve the filter eff iciency.

关 键 词：粒子滤波 重要性密度函数 重采样 高斯辅助粒子滤波 

分 类 号：TP24[自动化与计算机技术—检测技术与自动化装置]