抗差自适应Sage滤波及其在组合导航中的应用

Robust Adaptively Sage Filtering for Integrated Navigation System

针对卡尔曼滤波需要精确已知状态数学模型及其统计特性的问题,提出一种抗差自适应Sage滤波算法。该方法以Sage滤波为基本框架,吸收了抗差估计和自适应滤波的优点,利用Sage滤波开窗法求得观测残差向量和新息(预测残差)向量的协方差阵,由抗差估计方法确定观测噪声协方差矩阵,利用自适应因子调整动力学模型噪声协方差矩阵,以控制观测异常和动力学模型噪声对导航精度的影响。将提出的算法应用到捷联惯性导航(SINS)/合成孔径雷达(SAR)组合导航系统中,并与Kalman滤波和Sage滤波进行比较分析,仿真结果表明,提出的新算法不但能有效地控制观测异常和动态模型异常对状态参数估值的影响,而且能够抵制状态扰动,提高组合导航系统的滤波精度。

In order to solve the questions of Kalman filtering requires precise known state mathematical model and its statistical properties,a robust adaptively Sage filtering algorithm is presented,which absorbs the merits of robust estimation and adaptive filtering based on Sage filtering.Sage filtering window method is used to obtain covariance matrixs of observation residual vector and innovation（prediction residual） vector,and observation noise covariance matrix is determined by the robust estimation.Dynamic model noise covariance matrix is adjusted by using adaptive factor to control the influence of observation residuals and predict residual for navigation accuracy.The proposed algorithm is applied to SINS/SAR integrated navigation system.Experimental results and their comparison demonstrate that the proposed method can not only effectively control influences of observation and dynamic model anomaly on state parameter valuation,but also effectively resist the state disturbances,and navigation accuracy is much higher than Kalman filtering and Sage filtering.