摘要
针对微惯性测量单元精度较低及正交变换容积卡尔曼滤波(Transformed Cubature Kalman Filter,TCKF)协方差矩阵非正定导致滤波终止的问题,提出了一种基于奇异值分解的改进TCKF(Attitude Singular Value Decomposition Transformed Cubature Kalman Filter,ASVDTCKF)姿态估计算法。该算法以陀螺仪、加速度计和磁力计解算的姿态角作为量测,以姿态四元数作为状态,采用奇异值分解代替Cholesky分解的ASVDTCKF进行数据融合,实时估计姿态角。实验结果表明,在静态和动态条件下,ASVDTCKF能够有效解决TCKF误差协方差矩阵非正定造成的滤波终止问题,与平方根TCKF相比,姿态估计精度和运算效率平均提高了10.6%和19.7%,更适用于矿山等复杂环境的载体姿态测量。
In view of the low accuracy of the micro inertial measurement unit and the filter termination problem caused by the non-positive determination of the covariance matrix of transformed cubature Kalman filter,an attitude estimation algorithm of improved TCKF(attitude singular value decomposition transformed cubature Kalman filter,ASVDTCKF)based on singular value decomposition(SVD)was proposed.In this algorithm,the attitude angle calculated by the gyroscope,the accelerometer and magnetometer was used as measurement,the attitude quaternion was used as state,and ASVDTCKF,which used singular value decomposition instead of Cholesky decomposition,was used for the data fusion to estimate attitude angle in real time.Experimental results showed that ASVDTCKF could solve effectively the filtering termination problem caused by the non-positive determination of TCKF error covariance matrix under the static and dynamic conditions.Compared with the square root TCKF,the attitude estimation accuracy and the calculation efficiency increased by 10.6%and 19.7%on average,which was more suitable for the carrier attitude measurement in complex environments such as the mines and others.
作者
钱镭源
张建强
许国强
曹文涛
Qian Leiyuan;Zhang Jianqiang;Xu Guoqiang;Cao Wentao(Institute of Intelligent Manufacturing and Intelligence Mining,Yuncheng Vocational and Technical University,Yuncheng 044000,China;Yangmei Fengxi Fertilizer Industry(Group)Co.,Ltd.,Yuncheng 044000,China)
出处
《矿山测量》
2021年第6期32-38,共7页
Mine Surveying
基金
高等学校科技创新项目(201701D31111184)。
关键词
姿态估计
微惯性测量单元
正交变换容积卡尔曼滤波
协方差矩阵
奇异值分解
attitude estimation
micro inertial measurement unit
transformed cubature Kalman filter
covariance matrix
singular value decomposition
