利用LS_AR模型对极移参数的中长期预报

Improvement of LS_AR Model and Long-term Forecast of Polar Motion

为了满足深空探测器自主导航定位对极移参数中长期预报的需求,阐述了基于LS_AR模型的极移参数中长期预报和精度评定的原理,提出了4种改进方案对LS_AR模型的构建进行优化,并利用IERS提供的1990～1996年的极移参数的时间序列检验4种优化方案,得到了最优的预报模型,在400 d跨度上,其预报结果的平均绝对误差比未优化的模型小3 mas左右。

In order to satisfy the middle and long term forecast demand of polar motion parameters for navigation and positioning of the deep space probe,in this paper,the theory of polar motion parameter for middle and long term prediction and accuracy evaluation with LS_AR model are presented. Four methods to improve LS_AR model are put forward. And the best method is selected by checking them with polar motion parameter time series from 1990 to 1996 provided by IERS. The prediction result indicates that MAE of the most optimal prediction model＇s solutions is about 3 mas less than that of the unoptimized model for 400-day span.