基于GM(1,1)+AR模型的钟差短期预报改进算法研究

针对传统GM(1,1)+AR组合模型的缺点，提出一种可及时更新建模序列和增强数据间相关性的循环式钟差预报模型，在预报过程中根据预报时刻的不同实时调整AR模型阶数。考虑到原始钟差建模序列长度会对预报精度造成影响，分别使用2 h、6 h、12 h和24 h的钟差序列构建模型。实验结果表明，改进模型的预报精度较传统方法有一定提高，且预报结果更稳定；使用不同长度的钟差序列构建模型对预报结果有一定影响，其中二次多项式模型受原始序列长度的影响较大，改进模型受影响较小。

Research on the Improved Algorithm of Clock Bias Short-Term Prediction Based on GM(1,1)+AR Model

Considering the shortcomings of the traditional GM(1,1)+AR combination model, we propose a cyclic clock bias prediction model, which can update the modeling sequence in time and enhance the correlation between the data. The order of AR model is adjusted in real time according to different forecast times. Considering the influence of the original clock bias modeling sequence length on prediction accuracy, the clock bias sequences of 2 h, 6 h, 12 h and 24 h are used to build the model respectively. The results show that the prediction accuracy of the improved model is superior to the traditional method, and the prediction results are more stable. Using clock bias series with different lengths to build the model will have a certain impact on the prediction results, among which the quadratic polynomial model is relatively more affected by the length of the original series, and the improved model is relatively less affected.