中文摘要：

首先给出扩展卡尔曼滤波（Extended Kalman Filter，EKF）的原理，通过分析粗差在EKF模型中传递特性，给出新的抗差EKF模型。模型根据多余观测分量及预测残差统计，构造抗差等价增益矩阵，通过迭带给出GNSS抗差导航解。为提高模型在动态导航应用中的效率，文章结合统计模型，仅对存在粗差的观测历元进行抗差估计，进一步提高模型实时运行效率。并模拟GPS／Galileo多卫星导航星座及接收机平台的动态轨迹。采用加速度导航方程验证本文模型，并对不同模型运行的时间进行比较。结果表明在粗差存在的情况下，本文模型仍能正确导航，并且改进后的模型能明显提高实时导航的效率。

英文摘要：

The Extended Kalman Filter（EKF） principal is firstly investigated. A new robust EKF model is proposed based on the effect feature of the outliers on EKF. The proposed model implements equivalent Kalman gain matrix built by introducing redundancy and predicted residuals. The iterative scheme is suggested for solving the GNSS robust EKF solution. For improving the efficiency of the real time navigation, the model is further improved by combining statistic model and only the epoch with outliers is given solution with robust EKF. The GPS/Gallieo combination constellations are used to simulate a 11-state GNSS navigation case. A dynamic moving re- ceiver trajectory is designed to test the new filter models and the time needed is compared. Simulation results show that the suggested algorithm can gives correct navigation while there are outliers and the improved robust EKF is faster.