摘要
针对系统噪声和量测噪声均发生变化时,现有自适应卡尔曼滤波容易发散导致SINS/GNSS导航精度下降的问题,提出了一种基于Allan方差的改进自适应滤波SINS/GNSS导航算法。该方法在对自适应滤波进行改进的基础上,结合Allan方差估计法计算量测噪声协方差阵,克服了自适应滤波中噪声参数耦合以及高维度系统出现奇异性导致滤波发散问题,并利用残差χ2故障检测法对系统状态进行判断,对遗忘因子进行动态调整,对噪声特性跟踪效果更快速,相比其他改进方法简单易实现。仿真结果表明,与卡尔曼滤波,Sage-Husa自适应滤波相比,所提出的算法对噪声有较好的估计效果,且导航精度更高,滤波稳定性更好,速度均方误差平均可比传统Kalman滤波提高49.06%,较Sage-Husa自适应滤波提高27.19%;位置均方误差平均可比传统Kalman滤波提高41.12%,较Sage-Husa自适应滤波提高19.79%。
In view of the problem that SINS/GNSS navigation accuracy degrades due to divergence of existing adaptive Kalman filter when both state noise and measurement noise change.An improved adaptive filtering algorithm based on Allan variance is proposed.On the basis of the improvement of adaptive filtering,Combined with the Allan variance estimation method,the measurement noise covariance matrix is calculated,overcoming the problem of noise parameter coupling and filtering divergence caused by singularity in high-dimensional systems in adaptive filtering,and on the basis of judging the system state by using fault detection method,the forgetting factor is dynamically adjusted to track the noise characteristics more quickly.Compared to other improvement methods,it is simple and easy to be implemented.The average speed mean square error can be increased by 49.06% compared to traditional Kalman and 27.19% compared to Sage-Husa adaptive filtering.The average position mean square error can be increased by 41.12% compared to traditional Kalman and 19.79% compared to Sage-Husa adaptive filtering.
作者
祁帅
贾继超
寇得民
刘鑫
牟含笑
Qi Shuai;Jia Jichao;Kou Demin;Liu Xin;Mou Hanxiao(16th Institute,China Aerospace Science and Technology Corporation,Xi’an Shaanxi 710100,China)
出处
《传感技术学报》
CAS
CSCD
北大核心
2024年第5期818-824,共7页
Chinese Journal of Sensors and Actuators
关键词
组合导航
改进自适应滤波
ALLAN方差
遗忘因子
integrated navigation
improved adaptive
Kalman filter
Allan variance fading factor
