基于时间序列分解的GPS卫星钟差预报

GPS satellite clock error predicting based on ARMA analysis

根据二次多项式和灰色模型在卫星钟差预报中的特点,本文提出以ARMA(Autoregressive Moving Average)模型分别进行改进;对IGS(International GPS Service for Geodynam ic)提供的精密钟差进行时间序列分解,根据预报时间的长短,趋势项分别采用二次多项式和灰色模型拟合,随机项采用ARMA模型进行模拟,构建出了钟差的时间序列模型,并对未来的钟差进行了短期和长期预测,算例表明,均取得了良好的结果。

According to the benefits of quadratic polynomials and the Gray model in prediction of satellite clock error,this paper proposed to improve it with ARMA（Autoregressive Moving Average） model.To decompose the precise GPS clock error provided by the IGS（International GPS Service for Geodynamic）,the trend item was fit in accordance with quadratic polynomial model and the Gray model whose difference lies only in the length of time,and the random item was simulated by ARMA model.At last,in the short-term and long-term GPS clock error forecasts with ARMA,good results were achieved.