一种顾及钟差周期误差和随机特性的卫星钟差预报方法

提出一种顾及钟差周期误差和随机特性的卫星钟差预报方法。首先通过比较二次多项式加1、2、3、4个主要周期误差的模型，取其优者求得钟差预报的拟合值;然后针对拟合残差值的随机特性采用灰色模型进行建模，求得拟合值残差预报值；最后，将其与之前求得的预报值相结合得到最终的钟差预报值。采用IGS的15 min精密钟差数据进行实验，结果表明，在短期预报中，加2个主要周期误差的模型预报性能最好，并且新模型的预报精度优于常用算法。

A Method of Satellite Clock Bias Prediction Considering Periodic Errors and Stochastic Characteristics

In order to improve satellite clock bias prediction, a new prediction method is proposed considering periodic errors and stochastic characteristics. First, the given satellite clock bias is fitted by four quadratic polynomial models with one to four dominating periodic errors; the best model is then chosen to obtain the fitting residuals. Then, the prediction of the fitting residuals is modeled based on grey model, considering the stochastic characteristics of the fitting residuals. Finally, the clock bias based on the best of the four models and prediction of the fitting residuals are combined to obtain the ultimate prediction result. The precise data of satellite clock bias within 15 min from IGS are used to conduct experiments on different models. The results show that the model with two dominating periodic errors is better than the model with other dominating periodic errors and that the proposed model performs better than commonly used models in short-term prediction.