摘要
针对卡尔曼滤波在系统模型失配和未知干扰情况下鲁棒性差的特点,对于一类线性模型提出了多渐消因子卡尔曼滤波算法。该算法利用卡尔曼滤波取得最佳增益时残差序列互不相关的性质,可以在线自适应地调整多个渐消因子,从而对多个数据通道进行渐消,即使当滤波达到稳态时仍然可以调整滤波增益,使得该算法对模型失配和未知干扰有较强的鲁棒性。将该算法用于噪声统计不准确的SINS初始对准,数值仿真表明,当系统模型存在不准确情况时,新方法对航向误差角的估计精度较单渐消因子卡尔曼滤波和常规卡尔曼滤波分别提高了70%和43%,证明了该算法的有效性。
In view of poor robustness of Kalman filter algorithm when there exits model errors or unknown external disturbances,a multiple fading factors Kalman filter is proposed for a kind of linear model,which is based on the property that the sequence of residuals is uncorrelated when Kalman filter gain is optimal.The new algorithm can adjust multiple fading factors adaptively,therefore it can fade multiple data channels.When the filter becomes steady,it can also adjust the filter gain,which makes it have strong robustness when there exits model errors or unknown external disturbances.The proposed algorithm is applied to strapdown inertial navigation system（SINS） initial alignment with inaccuracy noise statistics,and simulation indicates that,compared with those of single fading factor Kalman filter and conventional Kalman filter when system model is inaccuracy,the estimating accuracy of heading error angle of the proposed algorithm is improved by 70% and 43% respectively,which proves the effectiveness of the new algorithm.
出处
《中国惯性技术学报》
EI
CSCD
北大核心
2012年第3期287-291,共5页
Journal of Chinese Inertial Technology
基金
国家自然科学基金(61104036)
关键词
卡尔曼滤波
单渐消因子
多渐消因子
噪声统计不准确
鲁棒性
Kalman filter
single fading factor
multiple fading factors
inaccuracy noise statistics
robustness
