中文摘要：

针对导航星座自主定轨,提出一种提高集中式算法效率的新思路,即充分利用高频、高精度的星间链路测距信息,在短弧内将卫星最优轨道与长期预报轨道的差异用多次曲线描述,得到卫星位置和速度的最佳估值。此方法无需动力学建模和计算状态转移矩阵,因此算法极为简洁。同时,对于自主运行期间缺乏空间基准,提出约束轨道升交点赤经的方法,以减小对地面系统的依赖程度。仿真结果表明,导航星座自主运行60 d,不考虑地球自转参数（EOP）长期预报误差,在无锚固站的情况下,链路数不少于5条时能够达到轨道URE优于1 m,位置3 m,速度毫米级的定轨精度。最后,通过比对验证了新算法比已有EKF分布式自主定轨算法的效率更高。

英文摘要：

A new rapid centralized algorithm is proposed to improve the low computational efficiency for autonomous orbit determination of navigation satellite constellation, which fully uses the powerful high frequency inter-satellite measurement capability, and describes the difference between the real and long term forecasting orbit as the simple higher order polynomials in short arc, and then obtains the satellite＇ s optimal position and velocity estimation. This method needs no complex dynamic modeling, trajectory integration and the state transition matrix calculation, and thus greatly saves computing resources. Meanwhile in order to reduce the dependence on ground support system, the constraint of the longitude of the ascending node information of forecast orbit is proposed to establish the spatial datum. The simulation results show that under the existing simulation scenarios after 60 days autonomous operation the new algorithm has the orbit user orbit error（URE） of 1 m , the position error of 3 m, and the velocity error of mm level with no less than 5 inter-satellite links, regardless of EOP error. Finally, by comparison, the autonomous orbit determination time using the new method is more efficient than the distributed Extended Kalman Filter （EKF） algorithm.