改进的多项式+周期项模型的卫星钟差预报

针对现有的超快速钟差产品IGU精度较低以及无法满足实时PPP技术的问题,提出了一种改进的多项式+周期项钟差预报模型。该模型采用多项式+周期项非线性函数对钟差数据进行滑动估计,结合迭代法对拟合模型的随机误差进行自然修正,以实现对卫星钟差的预报估计。通过与常见的多项式模型、灰色系统模型和多项式+周期项模型的对比分析,结果表明:改进的多项式+周期项模型更加适用于卫星钟差预报,在1天内,其预报精度RMS可以达到0.57 ns,最大偏离程度为1 ns,明显优于灰色系统模型和多项式+周期项模型;随着预报时间的增长,多项式模型、灰色系统模型和多项式+周期项模型的预报精度大幅降低,而改进的多项式+周期项模型没有大幅的变化,预报结果比较稳定。 

Satellite Clock Error Prediction of Improved Polynomial and Periodic Mode

In order to solve the problem that accuracy of the existing ultra-fast clock error products is too low to meet the real-time PPP technology, an improved polynomial and periodic clock error prediction model is proposed. The model first uses the polynomial and periodic nonlinear function to make a sliding estimation of the clock error data, and then uses the iterative method to naturally correct the random error of the fitting model to realize the prediction and estimation of the satellite clock error. Compared with the common polynomial model, the gray system model and the polynomial and periodic model, the results show that the improved polynomial and periodic model is more suitable for satellite clock error forecasting, and RMS of the forecast results can reach 0.57 ns and the maximum deviation is 1 ns within a day, which is obviously better than the gray system model and the polynomial and periodic model. With the increase of forecasting time, the forecasting accuracy of the polynomial model, the gray system model and the polynomial and periodic model greatly decreases, while the improved polynomial and periodic model does not change significantly, and the forecast result is stable.