In order to improve the operation efficiency of airport traffic, a prediction model of departure and departure time based on BP neural network was established. The key factors influencing the actual slip out time were analyzed, and the correlation was tested by SPSS. The model was verified by two weeks" actual operation data of an airport in central and southern China, and the root mean square error (RMSE), mean absolute error (MAE) and mean absolute error percentage (MAE) were analyzed. The results show that the new quantifiable factor, that is, the mean taxi time within one hour, has a significant effect on the actual taxi time. Although the time of the flight can also be quantified, the accuracy of the prediction is significantly reduced when this factor is added. The error between the prediction and the actual value was less than 5min, accounting for 84%.