本文基于时间序列分析的DDS(Dynamic Data System)建模法,对季节性的电离层总电子含量(TEC)时间序列观测值平稳化后建立自回归AR模型,提出以半参数AR模型对普通AR模型精化,并利用半参数AR模型对电离层TEC预报。实例分析表明:利用半参数AR模型对电离层TEC进行预报,在短期内半参数模型预报效果优于普通AR模型,但随着预报时间变长,则半参数模型预报精度明显下降,其预报效果则不如普通的AR模型。 In this paper, based on DDS (Dynamic Data System) modeling method of time series analysis, the autoregressive AR model is established after stabilizing the seasonal time series observations of ionospheric total electron content (TEC), and the semi-parametric AR model is proposed to normal AR The model is refined and the ionospheric TEC is predicted using a semi-parametric AR model. The case study shows that semi-parametric AR model is used to predict ionospheric TEC. The prediction of semi-parametric model in short-term is superior to that of ordinary AR model. However, as the forecasting time increases, the accuracy of semi-parametric model prediction decreases obviously. It is not as good as a normal AR model.