基于多项式模型的改进卫星钟差预报方法

The Optimization Method for the Satellite Clock Error Prediction Based on Polynomial Model

根据卫星钟固有的特性,提出了一种通用的卫星钟差预报方法。该方法在多项式模型的基础上,对多项式模型的选择、周期项修正、残差建模等方面进行了改进。首先利用拟合数据的残差标准差指标进行多项式模型的自适应选择,继而利用频谱分析的方法对钟差序列的周期项进行了建模,最后利用AR模型对消除趋势项的随机项进行建模。试验结果表明,改进的钟差预报模型能够显著地提高钟差的预报精度,对于稳定度较差的卫星的预报效果尤为显著。改进模型的卫星钟预报6h预报精度总体优于1ns,24h预报精度优于2ns.

According to the physical characteristics of satellite clock,this paper adopts a universal clock offset model based on polynomial prediction model.Multiple improvements have been verified,which contains the self-adaptive selection of polynomial model,modeling for periodic term and modeling for the residual term.Firstly,the adaptive selection of polynomial fitting model is judged by the standard deviation of residual.Then the periodical term of clock offset is modeled using spectrum analysis.Finally the stochastic model after removing the periodical term is constructed using AR model.The experiment results show that the proposed model can greatly improve the short-term prediction accuracy,especially for less stable satellite clock.The overall accuracy of 6hprediction can be better than 1ns and24 hprediction accuracy better than 2ns.