The new generation of information technology has had a significant impact on people’s behavior, and making airports more intelligent and intelligent. Airport intelligence is an important component of smart cities and intelligent transportation. Accurate prediction of passenger throughput is not only one of the keys to achieving a smart airport, but also of great significance for airport management and operation. The artificial intelligence prediction method SVR has advantages such as fast convergence speed and suitability for small sample data. The BA algorithm has advantages in parameter optimization, and the combination of the two can effectively improve prediction performance. This article takes Shuangliu Airport in Sichuan Province, China as the research object. In response to the nonlinear characteristics of the airport’s monthly passenger throughput from 2001 to 2019, based on the analysis of the correlation between passenger throughput with historical data and takeoff and landing sorties, a prediction model SVR-BA was established using SVR and BA, while BP-BA and ARMA models were also established. The prediction performance of several types of models was compared from MAPE and MAE, and it was found that the SVR-BA model had better prediction performance and higher robustness. This indicates that after optimizing the parameters of the BA algorithm, the predictive performance of the SVR prediction model can be effectively improved.