摘要
在基于卡尔曼滤波的捷联惯导初始对准中,由于噪声协方差矩阵无法精确确定,滤波过程中状态估计协方差矩阵的数值稳定性不好,容易导致滤波精度降低和状态估计协方差矩阵失去对称正定性,甚至滤波发散。针对这一问题,提出将UD分解自适应卡尔曼滤波应用到捷联惯导初始对准问题中。新算法采用协方差匹配法,实时估计并调整噪声协方差矩阵,对状态协方差矩阵的分解矩阵更新,保证协方差矩阵的对称正定性。仿真结果表明,相比标准卡尔曼滤波,UD分解自适应滤波算法数值稳定性好,具有噪声自适应能力,在先验量测噪声协方差矩阵存在误差的情况下,新算法能缩短初始对准的时间。
During the fine alignment process of inertial navigation system,Kalman filter is usually used to estimate attitude errorangles. Due to the reason that standard Kalman filter's numerical stability is not good and a priori measurement noise covariance matrix is difficult to accurately determine, the adaptive Kalman filter based on UD factorization and stochastic model covariance matching method is proposed to inhibit the calculation divergence of Kalman filter and improve the algorithm's measurement noise adaptive capacity. Numerical simulation results show that when priori covariance matrix has an error, compared to the standard Kalman filter, the new algorithm improves the accuracy and shorten the time of the initial alignment.
出处
《计算机工程与设计》
CSCD
北大核心
2014年第1期158-162,212,共6页
Computer Engineering and Design
关键词
初始对准
卡尔曼滤波
分解滤波
协方差匹配法
捷联惯导系统
initial alignment
Kalman filter
factorized filtering
eovariance-matching method
strapdown inertial navigation system