In order to more accurately explore the trend of the number of people studying abroad in China, a combination forecasting model based on L1 norm is proposed to make a prediction about the number of people studying abroad. The factors that affect the number of students studying abroad are selected from multiple perspectives, and the typical factors that affect the number of students studying abroad are extracted by grey correlation analysis, and the GM(1,3) model is constructed. BP neural network model is established. Combination prediction model based on L1 norm is proposed, and the optimal weight coefficient of single model is determined by solving linear programming. Subsequently, it forecasts the number of students studying abroad from 2006 to 2019. The GM(1,1) model is selected as the control model, and the prediction accuracy of the model is compared through the control model and the prediction error evaluation index system. The results show that the combination forecasting model based on L1 norm outperforms three single models, which effectively improves the prediction accuracy and makes full use of the information provided by the single prediction model, thus more accurately predicting the number of students studying abroad. In the next few years, there is still a large space for the development of the scale of studying abroad in China. The prediction results can provide reference for the relevant work of studying abroad in China under the global epidemic situation.