In recent years, the civil aviation transportation industry is developing rapidly. The airport has become an indispensable facility for passenger travel and cargo transportation. With the continuous increase of airport throughput, traffic congestion often occurs during flight peak hours and bad weather, resulting in large-area flight delays and long-time ground waiting of aircraft, especially those flights that have entered the taxiway after launch, resulting in high aviation fuel loss with the operation of APU (auxiliary power unit). Therefore, the management of flight delay has become an effective means to improve the competitiveness of air transportation enterprises. This paper studies the number of aircraft taking off and landing and the law of time difference on each runway of the airport. Taking the time difference as the independent variable, through fitting the data, this paper establishes the relationship model between the take-off and landing difference and the airport operation efficiency, completes the model hypothesis, model establishment and stability analysis, tests the goodness of fit of the actual flight data of the airport, takes the test results as the model input for simulation, and the results are consistent with the actual departure process. Finally, through the analysis of departure flight data, this paper uses the method of delaying the number of flights in peak hours in batches to improve the algorithm and improve the utilization of taxiway and runway, so as to reduce the waiting time and dwell time of flights. Through the simulation of the improved model, the results show that the improved algorithm is practical and effective, and provides an effective reference value for solving the problem of airport ground traffic congestion.