GPS/DR组合导航抗差自适应滤波算法  

Adaptive robust filtering algorithm in GPS/DR integrated navigation

作　　者：焦雅林[1] 高社生[1] 薛丽[1] 

机构地区：[1]西北工业大学自动化学院,西安710072

出　　处：《中国惯性技术学报》2010年第3期307-311,共5页Journal of Chinese Inertial Technology

基　　金：航空科学基金(20080818004);陕西省自然科学基金项目(N9YU0001)

摘　　要：针对GPS/DR车辆组合导航系统的数学模型具有非线性,应用扩展卡尔曼滤波进行线性化会导致滤波结果出现较大误差的问题,引入了抗差自适应滤波算法。利用计算机仿真,分别对抗差自适应Kalman滤波和自适应Kalman滤波算法进行仿真验证,结果表明,抗差自适应滤波不但能自适应地确定观测噪声的协方差矩阵,而且能利用自适应因子调节状态参数噪声的协方差矩阵,可以控制观测异常和动态模型噪声异常对状态参数估值的影响,使状态参数的估值更加合理。自适应Kalman滤波使位置误差控制在30m,而对抗差自适应Kalman滤波能使位置误差控制在18m左右,且误差控制更稳定。In view that the mathematical model based on GPS /DR vehicle integrated system is non-linear,and using EKF（extended Kalman filter）to make linearization could cause significant errors in its filtering results,the paper introduces an adaptive robust filter.The simulative computation of the adaptive robust filtering and adaptive Kalman filtering are studied respectively,and the results show that the robust adaptive filtering can adaptively determine the covariance matrix of observation noise,and can adaptively regulate the covariance matrix of state parameter noise by adaptive factor.This method can control the impacts of observed noise abnormal and dynamic model noise abnormal on the estimation of state parameters,and make the state estimation more reasonable.The experimental result shows that the adaptive robust filtering algorithm reduces the position error to 30 m;while the position error by the adaptive Kalman filtering algorithm reduces it to 18 m,and its error control is more stable.

关 键 词：GPS/DR组合导航 抗差自适应滤波 KALMAN滤波 航位推算 

分 类 号：U666.1[交通运输工程—船舶及航道工程]