Three kinds of algorithm and its performance analysis in the prediction of satellite clocks bias

It is a very key technology not only in the autonomous operation of the navigation satellite system but also in the optimizing the satellite clock bias parameters for navigation message that the high-precision and reliable prediction of in-orbit atomic clocks. In order to compare the performance of three models in the prediction of satellite clock bias, sixteen separate tests have been done with different long satellite clock bias data series to establish the three models, and the satellite clock bias data series in different time periods are predicted. The prediction results of three models were evaluated by using different evaluation criteria. The evaluation results show that linear polynomial prediction model, quadratic polynomial prediction model, and gray prediction model are basically the same in the short-term prediction of satellite clock bias. And the linear polynomial prediction model and gray prediction model are better than quadratic polynomial prediction model in the mid-long term prediction of satellite clock bias, and linear polynomial prediction model and gray prediction model are more suitable for the mid-long term prediction of satellite clock bias.