Research Of Predictability Of Departure Flow In Capital Airport Based On Time Series

In view of the problem of airport traffic prediction and control,this paper starts with the analysis of the nonlinear dynamic characteristics of the airport traffic flow,and studies the predictability of the airport traffic flow,which is of great significance to the prediction and control of the airport traffic flow.Firstly,this paper obtains the planned departure time and actual departure time by processing and collecting the FPL of BeiJing Capital Airport.According to the three cases of continuous day,working day and weekend,time series of 60 min,30min,15 min and 10 min are constructed.Phase space reconstruction is used to reconstruct the phase space by phase space reconstruction,and the best time delay and embedding dimension are calculated by C-C algorithm.The maximum Lyapunov exponent calculated by the small amount of data is calculated and the correlation dimension calculated by the G-P algorithm is used to determine the chaotic and fractal characteristics of the time series of the traffic flow.The results show that the time series of different time scales all have chaotic characteristics,and the chaotic characteristics of the time series under the 10 minute scale are the strongest.Secondly,the predictability of the sample data is analyzed.When the sample size increases,the predictability of the traffic flow will increase and the predictability attenu ation rate of the traffic flow will slow down.The effect of noise on predictability is analyzed.After smoothing the time series of the departure traffic flow,it can significantly improve the predictability of the traffic flow and increase the accuracy of the prediction.In the end,the time series of scheduled departure time series and actual departure time are compared.It is found that the chaotic characteristics of the planned departure time series are more weak than the time series of the actual departure time,but the predictability of the planned departure time series is stronger than the actual departure time series.The research results in this paper have important application value for reducing the running cost of airlines and airports and improving the service quality of passengers.