Abstract：To solve the problems that most ionospheric products are based on the grid form and the low accuracy of traditional forecast models, we use a weighting function method to calculate single-station TEC and propose a sliding window-based improved MEA-BP(MW-MEA-BP) forecast model. We select the IGS stations at different latitudes to analyze and validate the observations in low and high solar activity years, respectively. The results show that the proposed single-station TEC estimation method is fast and reliable, and the MEA-BP model is a significant improvement over the BP model in short-term forecasting. When the TEC changes drastically, the forecast accuracy of the MW-MEA-BP model improves 71.8%-90.6% compared with the BP model, and 13.9%-54.4% compared with the MEA-BP model. The research can provide a priori accurate ionospheric information for positioning users.