Research On Identification And Prediction Methods Of Air Traffic Congestion

The identification and prediction of air traffic congestion is the basis of air trafficcongestion management. At present, the researches of air traffic congestionidentification and prediction have not been formed as a scientific and theoreticalsystem. In this paper, this problem was deeply studied from the following aspects:analysis of congestion behavior; establishment of congestion indexes; congestionidentification and prediction of airports, crossing air routes and terminal areas;prediction of congestion propagation in airport network.(1)The concrete definition of air traffic congestion was provided based on inflowand outflow rates, which reflects the essence and dynamic progress of congestion. Thecongestion behavior was analyzed based on complex network theory from structureand function aspects: in order to mine out the necessary structure conditions ofcongestion, the evolution of flight network topology was analyzed; in order to mineout the factors influencing congestion, the fluctuation of traffic flow in flight networkwas analyzed. Then the congestion indexes were established. They were founded fromdifferent congestion characteristic of airports, crossing air routes and terminal areas:stranded degree index which is used to represent the movement feature of congestion;saturation index which is used to represent the load characteristic of congestion;aggregation degree index which is used to represent the complex characteristic ofcongestion; the amount of potential conflicts index which is used to represent theconflict feature of congestion.(2)The traffic congestion indexes of airports, crossing air routes and terminalareas were selected respectively based on stranded degree, traffic volume and theinternal characteristics of different airspace congestion. The identification method ofair traffic congestion based on gray cluster was proposed as the uncertainty andfuzziness of congestion. Numerical simulations were carried out which show that theidentification method can simulate human judgement on uncertainty concept well.(3)Prediction methods of different congestion indexes were proposed.Probabilistic prediction method of airport demand was established based on demanduncertainty. Prediction method of airport capacities was proposed based on PCA-Kalgorithms as the similarity of airport capacity. The aggregation model of crossing air routes was established from the aspect of chase aggregation and crossing aggregation.A multiple regression prediction model of aggregation index was also proposed. Thechaotic phenomena in flight conflict system were found through qualitative andquantitative analysis. A chaotic prediction method of potential conflicts wasestablished. Combination prediction method of stranded degree based on fuzzy softsets was proposed which combines statistical prediction method and intelligentprediction method, which can reflect the long-term periodicity and short-termnonlinear of trffic flow. Numerical simulations verified the validity of the above indexprediction methods.(4)Airport importance assessment model was established. It was used to identifybottleneck airports. The relative airports of bottleneck airports were identified basedon multi-dimensional scaling method. Prediction method of airport congestionpropagation based on Elman Neural Network was established, which was used topredict congestion situation of the relative airports when the bottleneck airport iscongestion. At last, numerical simulations were carried out which demonstrated thevalidity of the above methods.The dissertation researches the identification and prediction of air trafficcongestion deeply. The proposed methods have strong pertinence and are easy to berealized. It consummates the theory and methods of identification and prediction of airtraffic congestion.