ARM架构下的低轨卫星实时定轨精度及运算效率分析

Precision and efficiency analysis of real-time orbit determination for LEO satellites based on ARM architecture

针对基于ARM架构计算平台的低轨卫星实时定轨由于主频限制导致的运算效率瓶颈问题,旨在尽量不牺牲实时定轨精度的情况下,缩短对应实时窗口所需的预报时间,从而提高低轨卫星实时轨道精度及运算效率。本研究使用Sentinel-6A卫星星载GPS观测数据为例,结合法国国家空间研究中心(CNES)提供的GNSS实时精密产品,通过简化动力学方法在ARM架构服务器上进行基于批量最小二乘法的定轨及预报。研究分析了不同观测数据采样间隔、定轨弧长及拟合时长对定轨、预报轨道精度及运算效率的影响。结果表明,定轨弧长从24 h缩短至12 h的情况下,预报30 min轨道OURE及3D RMSE精度下降6.1%及5.9%左右,运算效率提高了45%~53%。相比于30 s采样间隔,延长采样间隔至120 s可使运算效率提高约41.72%,预报30 min轨道OURE精度下降2.56%左右,稳定在2到4 cm。在为轨道短期预报所测试的不同拟合时长中,当拟合时长为6 h,预报30 min轨道OURE精度最优,稳定在2~3.5 cm。

To address the computational efficiency bottleneck of Low Earth Orbit(LEO)Precise Orbit Determination(POD)on ARM-based platforms,this study aims to enhance the computational efficiency without significantly sacrificing orbit determination accuracy.In such a way,the prediction time required to catch up with the real-time time window can be shortened,and the accuracy of the real-time predicted orbits delivered to users can be improved.Using the GPS data on the LEO satellite Sentinel-6A as an example,this study used real-time GNSS products from the National Centre for Space Studies(CNES)in France,and performed LEO satellite POD and orbit prediction on ARM-based servers employing the reduced-dynamic method.The study analyzed the impacts of different observation sampling intervals,arc lengths and fitting times on the orbital accuracy and computational efficiency.Results showed that compared to a 24 h arc length,a 12 h arc degrades the OURE and 3D RMSE accuracy by approximately 6.1%and 5.9%,respectively,for predicted orbits of 30 min.However,the computational efficiency can be increased by 45%to 53%.When the sampling interval is 120 s,computational efficiency improves by approximately 41.72%compared to a 30 s interval,with the 30 min predicted orbital OURE decreased by approximately 2.56%,amounting to 2 and 4 cm.Among all the tested fitting times,the fitting time of 6 h leads to the best OURE for predicted orbits of 30 min,amounting to 2 to 3.5 cm.