摘要
GNSS/SINS(全球卫星导航/捷联惯性导航)组合导航系统的最优卡尔曼滤波需要满足测量值为高斯分布的假设条件,否则将会导致滤波精度下降甚至发散。鲁棒滤波是解决上述问题的有效方法,但在提高滤波精度方面存在一定的局限性。为了解决该问题,基于马氏距离和牛顿迭代法提出了一种改进的鲁棒滤波方法。首先,根据马氏距离判断测量值是否存在异常;然后,根据牛顿迭代法引入一个比例因子以对测量噪声协方差矩阵进行优化与调整,进而较精确地调整测量值与系统模型信息在滤波过程中的比重;最后,基于最优卡尔曼滤波(KF)、传统鲁棒卡尔曼滤波(RKF)及文中提出的改进鲁棒卡尔曼滤波(IRKF)对GNSS/SINS组合导航系统进行仿真分析及性能对比。实验结果表明,相对于KF,IRKF可提高位置及速度精度分别为14.1%及13.8%,相对于RKF,IRKF可提高位置及速度精度分别为8.1%及7.7%。
The optimal Kalman filtering of GNSS/SINS integrated navigation system needs to satisfy the assumption that the measurement vector is Gaussian distribution;otherwise,the filtering accuracy will decrease or even diverge.Robust filtering is an effective method to solve this problem,but it has certain limitations in improving the filtering accuracy.To solve this problem,this paper proposes an improved robust filtering method based on Mahalanobis distance and Newton iteration method.Firstly,the Mahalanobis distance is used to determine whether the measured value is abnormal.Secondly,a scaling factor is introduced to optimize and adjust the measurement noise covariance matrix according to Newton iteration method,and then adjust more precisely the proportion of measurement value and system model information in the filtering process.Finally,the GNSS/SINS integrated navigation system is simulated and analyzed based on optimal Kalman filter(KF),traditional robust Kalman filter(RKF)and improved robust Kalman filter(IRKF).The experimental results show that IRKF can improve the position and velocity accuracy by about 14.1%and 13.8%respectively when compared with KF,and can improve position and velocity accuracy by 8.1%and 7.7%respectively when compared with RKF.
作者
孙玉梅
王学伟
任宪洁
SUN Yumei;WANG Xuewei;REN Xianjie(School of Computer Science,Weifang University of Science and Technology,Weifang 262700,Shandong,China)
出处
《弹箭与制导学报》
北大核心
2024年第1期6-12,共7页
Journal of Projectiles,Rockets,Missiles and Guidance
基金
山东省高层次人才科研启动项目(KJRC2022009)资助。
关键词
组合导航系统
马氏距离
鲁棒滤波算法
比例因子
integrated navigation system
Mahalanobis distance
robust filtering algorithm
scaling factor
