Research On Traffic Situation Identification And Prediction Method In Multi-terminal Area

Terminal area is an important component set up in one or several major airports in air traffic system.So that terminal area’s operation directly decides the orderly operation of air traffic.Relevant traffic management personnel are required to correctly grasp the operating status and development trends of the terminal area traffic,so as to work out effective traffic management measures in a timely manner to support that the entire air traffic network runs smoothly.However,current researches on the identification and prediction method of traffic situation in terminal areas are only carried out on the basis of single terminal area by some scholars,which fail to take the mutual circulation relationship between terminal areas into consideration.To this end,this paper has researched on identification and prediction method of traffic situation towards multi-terminal areas.Based on current identification methods of traffic situation,this paper put forward an entropy method and genetic algorithm to improve the traffic situation identification method of FCM terminal areas.So that the FCM algorithm improved is able to globally search for the optimal initial clustering center during the clustering,and endow different weights to each index in accordance with the index importance of each clustering sample,which has improved the identification effect of the FCM algorithm to a greater extent.Compared with FCM algorithm,the Xie-Beni coefficient of the improved FCM clustering algorithm has been improved by 35%,indicating that the improved FCM algorithm is characterized in optimal clustering effectiveness.Meanwhile,through comparing the traffic trends of terminal areas in Shanghai and Chengdu,it has found that the improved FCM traffic situation identification model possesses optimal identification effectiveness compared with the previous FCM model.Based on current prediction methods of traffic situation,this paper proposed a terminal area traffic situation prediction method based on the GMNN(Graph Markov Neural Network)improved.Since the improved method is equipped with attention mechanism and inductive learning strategy based on the GMNN model,it not only can fully exploit the dependence of different terminal intervals,but possesses better generalization capabilities.In terms of the multi-terminal area traffic trend prediction tasks,the improved GMNN model has reached an accuracy rate of 90.6% on the test set,with a 7%,5.4%,2.3%,2.2% respective increase compared with BP(Back Propagation),GCN(Graph Convolutional Networks),GAT(Graph Attention Neural Network)and GMNN models.Ans the data visualization verification has indicated that the prediction results accord with objective laws.In addition,the improved GMNN model can achieve accurate prediction of traffic situation in multi-terminal areas.The multi-terminal area traffic trend identification and prediction method that is proposed in this paper can assist controllers at terminal area to make more reasonable surveillance decisions,and optimize the traffic operation of the terminal area at the micro level;while at the macro level,it can assist relevant traffic management personnel to establish more balanced air traffic management measures,so as to favor the coordinated development of the civil aviation transportation industry.