基于奇异谱分析的BDS卫星钟差周期项提取

Periodic term extraction of BDS satellite clock offset based on singular spectrum analysis

在建立星座自主时间基准时,必须扣除卫星钟差的周期波动,以避免将其引入系统时间。为准确地扣除周期波动,提出一种基于奇异谱分析(SSA)的BDS卫星钟差周期项提取方法。首先,经SSA分解获得钟差信号的多个重构成分;然后,引入重标极差分析方法计算各重构成分的Hurst指数,根据Hurst指数辨识钟差的低频信号主导分量和高频信号随机分量,从而实现周期项的重建提取。对不同类型的BDS卫星钟进行了分析和研究,结果表明:所提方法能够准确提取钟差周期项,所提取周期项的频谱特征比钟差多项式拟合残差的频谱更为清晰;利用多项式拟合方法扣除周期项后,万秒稳和日稳平均分别提升4.1%和2.4%,而利用所提方法扣除周期项后,万秒稳和日稳平均分别提升20.9%和34.1%,提高了卫星钟频率稳定度,为星座自主时间基准的建立提供了基础。

In establishing the autonomous time reference of the navigation constellation,the periodic fluctuations of satellite clock offset must be deducted to avoid introducing them into the system time.In order to accurately remove the periodic fluctuations,a decomposition,identification and rebuilt method based on singular spectrum analysis(SSA)is proposed to extract the periodic terms of BeiDou satellite clock offset.Firstly,multiple reconstructed components of the clock offset signal are obtained by SSA decomposition.Then,rescaled range analysis is introduced to calculate the Hurst index of each reconstructed component.The dominant low-frequency components and random high-frequency components of clock offset subsequently are identified according to the Hurst index,so as to realize the reconstruction and extraction of the period term.The analysis and study of different types of BDS satellite clocks show that the proposed method can accurately extract the period terms of satellite clock offset,and the spectral features of the extracted periodic terms are clearer than that of the polynomial fitting residuals.While the mean improvements are 4.1%and 2.4%respectively for the frequency stability at the ten-thousand and daily sampling intervals via the polynomial fitting method,through the proposed method the improvements can reach up to 20.9%and 34.1%,respectively,which improves the frequency stability of the satellite clock and provides the reference for the establishment of an autonomous time reference for the navigation constellation.