GNSS/INS组合导航拓展卡尔曼滤波和粒子滤波算法对比  

Comparison of Extended Kalman Filter and Particle Filter in GNSS/INS Integrated Navigation

作　　者：张园 苗晓婷 丛丹姝 王睿[1] 刘儿兀[1] ZHANG Yuan;MIAO Xiaoting;CONG Danshu;WANG Rui;LIU Erwu(College of Electronic and Information Engineering,Tongji University,Shanghai 201804,China;SAIC Motor Corporation Limited,Shanghai 201804,China)

机构地区：[1]同济大学电子与信息工程学院,上海201804 [2]上汽集团软件中心智能驾舱部,上海201804

出　　处：《系统仿真技术》2020年第4期207-211,共5页System Simulation Technology

基　　金：上汽基金项目(1905)。

摘　　要：为了评估全球导航卫星系统(GNSS)/惯性导航系统(INS)中各种滤波算法在实际应用场景中的性能,针对滤波算法所估计的状态向量不同,组合导航的方式一般包括直接法和间接法,而GNSS/INS组合导航中数据融合的关键是滤波算法,因此选取间接法拓展卡尔曼滤波(EKF)和间接法粒子滤波(PF)分别用于GNSS/INS松组合导航,同时增加了直接法粒子滤波以增强实验对比效果。在仿真和实测环境下,对拓展卡尔曼滤波和粒子滤波两种非线性滤波算法进行对比。实验结果表明,间接法拓展卡尔曼滤波适用于弱非线性、非高斯情况下的GNSS/INS组合导航,定位精度较高;考虑到状态向量维数且尽可能减少粒子数,间接法粒子滤波(PF)可以选用拓展卡尔曼滤波(EKF)来提供更优的建议密度函数;直接法PF适用于状态方程非线性情况,对各导航子系统的输出参数进行滤波估计,但滤波精度不如间接法的EKF算法和PF算法。According to the different state vectors estimated by filtering algorithm,the integrated methods of integrated navigation generally include direct method and indirect method.The key of data fusion in GNSS/INS integrated navigation is filtering algorithm,so indirect extended Kalman filter(EKF) and indirect particle filter(PF) are selected for GNSS/INS loose integrated navigation respectively,and direct particle filter is added to enhance the experimental contrast effect. In the simulation and measurement environment,the extended Kalman filter and particle filter are compared.The experimental results show that the indirect Kalman filter is suitable for GNSS/INS integrated navigation under weak linear and non-Gaussian conditions,and the positioning accuracy is high;considering the dimension of state vector and reducing the number of particles as much as possible,the indirect PF can choose EKF to provide better recommended density function;the direct method pf is suitable for the nonlinear state equation to filter the output parameters of each navigation subsystem However,the filtering accuracy is not as good as the EKF algorithm and PF algorithm.

关 键 词：GNSS/INS 组合导航 卡尔曼滤波 粒子滤波 

分 类 号：V249.328[航空宇航科学与技术—飞行器设计]