基于卡尔曼滤波重构GRACE-FO姿态数据

Results from GRACE-FO attitude determination based on Kalman filter

重力卫星姿态数据通过非保守力转换和KBR(K-Band Ranging)天线相位中心改正影响时变重力场模型精度,因此如何获取高精度的姿态数据是原始载荷数据处理的重要研究内容.GRACE-FO(GRACE Follow-On)每颗卫星均安装了三颗星敏感器和一个惯性测量单元(IMU,Inertial Measurement Unit)测量卫星姿态数据.星敏感器对姿态的低频部分敏感,惯性测量单元对姿态的高频部分敏感,融合这两类数据可以获得高精度的姿态数据.为此,本文首先基于扩展Kalman滤波,以四元数和陀螺仪漂移参数为状态变量推导了一种新的姿态Kalman滤波融合算法.然后基于GRACE-FO Level-1A实测数据验证本文提出的姿态Kalman滤波算法是否可行,并分析对反演时变重力场模型精度的影响.对于GRACE-FO星敏感器数据处理而言,融合多个星敏感器数据可以抑制高频部分噪声的精度,特别对于三个星敏感器融合的精度要略高于两个星敏感器融合.对于星敏感器与IMU融合而言,本文解算的姿态数据充分融合了IMU测量的角速度信息,相较于官方机构精度至少提高了3倍,并在0.01~0.1 Hz上噪声水平要低一个量级左右.对时变重力场反演而言,从反演时变重力场模型的阶方差和等效水高上看,在任务初期姿态数据误差引起的时变重力场模型误差大约为5%,而进入任务稳定期时反演的重力场模型精度基本一致.目前姿态数据的精度对反演的时变重力场模型的精度影响很小,影响时变重力场模型的精度主要来源于其他误差.

The gravity satellite attitude affects the accuracy of the time-variable gravity field model through non-conservative force transformation and K-Band Ranging(KBR)antenna phase center correction.Therefore,how to obtain high-precision attitude data is an important research content of raw payload data processing.Each satellite of GRACE Follow-On(GRACE-FO)is equipped with three star cameras and an inertial measurement unit(IMU)to measure satellite attitude data.The star cameras are sensitive to the low frequency of the spacecraft attitude,the IMU is sensitive to the high frequency of the spacecraft attitude,and we can obtain a high-precision attitude data product by fusing these types of data.Thus,based on the extended Kalman filter,we first derive the attitude Kalman filter method with quaternion and gyroscope drift parameters as state variables.Then,the GRACE-FO Level-1A data is used to verify whether the attitude Kalman filter proposed in this paper is feasible,and the influence on the accuracy of the recovered time-varying gravity field model is analyzed.For star cameras data processing,the high-frequency noise can be reduced by combining multiple star cameras,especially the accuracy of the combing three star cameras data is slightly better than the results of the two star cameras.For the fusion of the star camera and IMU,the attitude data calculated in this paper completely integrate the angular velocity information measured by the IMU,the accuracy is at least 3 times higher than the official institution,and the noise level at 0.01~0.1 Hz is about an order of magnitude lower.For the recovered time-varying gravity field,from the perspective of the degree variance and equivalent water height,the error of the recovered time-varying gravity field model caused by the difference attitude data is about 5%at the early mission,and the accuracy of the recovered time-varying gravity field is basically the same when entering the stable period of the mission.At present,the accuracy of attitude data has little effect on the accuracy of the recovered gravity field model,and the restriction of the accuracy of the time-varying gravity field model mainly comes from other errors.