摘要
针对超宽带(UWB)测距异常值、传统滤波方法中动力学模型不精准、状态向量误差协方差阵非正定等问题,提出一种基于奇异值分解的抗差自适应容积卡尔曼滤波算法,并将其应用于UWB室内定位中:以标准容积卡尔曼滤波(CKF)算法为基础,利用残差向量构造抗差因子消除观测异常值对定位解的影响;利用自适应因子对整体模型误差进行实时调整和修正以提高滤波精度;同时用奇异值分解代替乔莱斯基(Cholesky)分解以进一步提高滤波的稳定性。实验结果表明,所提算法相比传统的扩展卡尔曼滤波(EKF)、无迹卡尔曼滤波(UKF)、CKF算法,能够进一步提高UWB系统的定位精度和抗干扰能力,定位最大误差由1.5 m降至0.3 m,均方根误差小于0.05 m。
Aiming at the problems of ultra-wide band(UWB)ranging outliers,the inaccuracy of dynamic model in traditional filtering methods,and the non-positive definite state vector error covariance matrix,a robust adaptive volume Kalman filter algorithm based on singular value decomposition was proposed and applied to UWB indoor positioning.Based on the standard cubature Kalman filter(CKF)algorithm,the residual vector is used to construct the robust factor to eliminate the influence of the observed outliers on the positioning solution,and the adaptive factor is used to adjust and correct the overall model error in real time to improve the filtering accuracy.At the same time,the singular value decomposition is used to replace Cholesky decomposition to further improve the filtering stability.Experimental results show that compared with traditional extended Kalman filter(EKF),unscented Kalman filter(UKF)and CKF algorithms,the proposed algorithm can improve the positioning accuracy and anti-jamming ability of UWB system better.The maximum positioning error is reduced from 1.5 m to 0.3 m,and the root mean square error is less than 0.05 m.
作者
高嵩
宋佳鹏
房穹
张熙为
GAO Song;SONG Jiapeng;FANG Qiong;ZHANG Xiwei(School of Geomatics,Liaoning Technical University,Fuxin,Liaoning 123000,China)
出处
《导航定位学报》
CSCD
2023年第1期142-147,共6页
Journal of Navigation and Positioning
基金
国家自然科学基金项目(42074012)
辽宁省重点研发计划项目(2020JH2/10100044)
辽宁省自然科学基金计划指导计划项目(2019-ZD-0051)
辽宁省教育厅基础研究项目(LJ2020JCL016)。
关键词
超宽带(UWB)定位
奇异值分解
容积卡尔曼滤波
测距异常值
系统噪声
ultra-wide band(UWB)positioning
singular value decomposition
cubature Kalman filter(CKF)
ranging outliers
system noise
