基于抗差估计的全姿态多位置寻北

为解决寻北过程中需要人工调平而耗时较长的问题，结合全姿态多位置寻北模型，采用总体最小二乘法估算北向角。该算法将陀螺漂移转化为测量误差，而转台倾斜误差、轴系误差及安装误差等同系数矩阵误差，在忽略较少误差项的同时完成寻北过程；并根据实际数据中含有粗差的情况，提出抗差总体最小二乘方法，引入 等价权函数消除粗差数据的影响，从而提高算法在含有粗差数据时的寻北精度。仿真数据分析和实验结果表明：该算法能有效削弱粗差对寻北精度的影响，实现全姿态高精度寻北。

Multi-position North-Seeking at All-attitude Based on Robust Estimation

To solve the problem of North-seeking that need artificial leveling and time-consuming issue, the total least squares estimate north Angle is adopted, combined with the Multi-position North-seeking model. To ignore less error terms of in the North-seeking process, gyroscopic drift can be converted to measurement error, the tilt error , the shafting error and setting error can equal to coefficient matrix error. According to the actual data contains outlier, the Robust Total Least Squares method is proposed. Introducing the equivalent weight function to eliminate the influence of gross error data, thus find the precision of algorithm. The results of emulation and experimental examination show that the algorithm can weaken the effect of outlier and can realize the whole attitude to find north.