基于容积卡尔曼滤波的高斯粒子滤波算法  

Gaussian Particle Filter Based on the Cubature Kalman Filter

作　　者：赵丹丹[1] 刘静娜[1] 贺康建 ZHAO Dan-dan LIU Jing-na HE Kang-jian(School of Computer Science, Shaanxi Normal University, Xi＇an,Shannxi 710062,China School of information,Yunnan University, Kunming,Yunnan 650500,China)

机构地区：[1]陕西师范大学计算机科学学院,陕西西安710119 [2]云南大学信息学院,云南昆明650500

出　　处：《计算技术与自动化》2017年第1期82-86,共5页Computing Technology and Automation

摘　　要：高斯粒子滤波是一种免重采样的粒子滤波,不会出现粒子退化,但其重要性密度函数由于没有考虑到最新量测信息,使得滤波性能明显下降,且该算法没有较高的实时性。针对这个问题提出一种基于CKF的高斯粒子滤波算法—CKGPF算法。该算法利用CKF算法构造高斯粒子滤波的重要性密度函数,且在时间更新阶段借助CKF算法来完成只对高斯分布参数的更新。仿真结果表明,CKGPF算法相比于标准GPF算法不仅提高了滤波精度,而且还具有较好的实时性。Gaussian particle filtering is a kind of particle filtering without particle resampling, but its importance density function because there is no consideration to the latest measurement information, make the filter performance is significantty reduced,and the algorithm does not have good real-time performance. For this, a new Gaussian particle filter algorithm based on CKF is proposed. The importance density function of Gaussian particle filter is structured by using CKF, and the update of Gouss distribution parametors were completed by using CKF in the time update stage. The simulation results show that CKGPF algorithm not only improves the filtering accuracy, but also has better real-time, compared with the standard GPF.

关 键 词：高斯粒子滤波 重要性密度函数 实时性 容积卡尔曼滤波 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]