Research On Short Term Forecasting Method Of Air Traffic Flow

The short term prediction of air traffic flow provides the auxiliary decision-making information for the optimization of air traffic flow control and management.It has a guiding effect on the effectiveness,optimization and accuracy of decision-making.It is a key basic scientific problem that needs to be solved in air traffic flow management.This problem has been studied in this paper.Firstly,a short-term flow prediction method based on AFTN messages is proposed.According to the contents of messages,a string segmentation matching algorithm based on regular expression and a vertical profile model of aircraft based on BADA are proposed.The real message data is taken as an example.The result shows that the traffic prediction method based on AFTN message can reflect the trend of air traffic flow in airspace,but the average absolute error is about 1.5 sorties compared with the measured flow data.Secondly,the airspace sector traffic time series is constructed by handling radar trajectory data.The chaotic characteristics of the four experimental flow time series are identified from the qualitative and quantitative analysis.Aiming at the timeliness problem of chaotic characteristic identification for flow time series,a dynamic identification method of chaotic characteristics of flow time series is proposed.The results show that the flow time series gradually has chaotic characteristics as the data increases.Finally,a multi-step prediction algorithm for air traffic flow time series based on echo state network is proposed.The experiment result shows that the average absolute error of multi-step prediction accuracy of traffic time series based on echo state network is about 0.9sorties.The time series at different temporal scales have chaotic characteristics.The higher the temporal scale,the higher the accuracy of the echo state network prediction.The optimal setting range of the reservoir parameters are given when the echo state network is used to predict the flow at different temporal scales.This paper proposes a short-term flow prediction method based on AFTN messages firstly.Then,on the basis of systematically identifying the chaotic characteristics of the flow time series,a multi-step prediction method of flow time series based on echo state network is proposed.And then puts forward some theoretical basis for air traffic flow management.