Abstract: The innovation of the Unscented Kalman Filter（UKF） is often damaged because of the presence of outliers in the dynamic positioning, which make the filter no longer accurate or even divergent. To solve this problem, the principle of the UKF algorithm and the effect of outliers on UKF algorithm have been analyzed. An outlier rejecting UKF algorithm has been proposed. This algorithm combines the UKF algorithm with the eliminate of outliers, adjusts the gain factor or performs outliers rejecting calculation according to monitoring innovation, such that the UKF algorithm still be optimal when there are outliers in measurement. The result of simulation demonstrates that the algorithm can eliminate the bad effect of outliers on filter effectively and improve the measurement accuracy．

Key words: positioning, Unscented Kalman Filter（UKF）, outlier, innovation, gain, accuracy

摘要： 动态定位中野值的存在，使无迹卡尔曼滤波UKF（Unscented Kalman Filter）的结果不再准确甚至发散。针对这一问题，提出了一种具有抗野值性能的UKF算法。该算法将经典UKF算法与野值的剔除相结合，通过对新息序列的判断，对野值点进行处理，实时地调整滤波增益或者进行野值计算，使UKF算法在野值干扰下仍为最优估计。仿真证明该算法可以有效地辨识和剔除野值的干扰，抑制滤波的发散，提高了定位的精度。

关键词: 定位, 无迹卡尔曼滤波, 野值, 新息, 增益, 精度