Research Of The Taxi Time Prediction Method For Arrived Flights

To predict the taxi time of airport accurately,is one of the measures for reducing delay time,decreasing delay spread and expanding the capacity of airport surface traffic.The research of the taxi time prediction method for arrived flights in this paper,provides a reference for the next research of time arrangement for flight guarantee service,departure flight time plan.In this paper,the taxiing-route of arrived flights is designed by ant colony algorithm(ACA),then the ACA is improved adaptively to optimize the designed route.According to the relevant data which is collected by Adaptive ACA,the subsequent taxi time on the nodes of the designed route is predicted by nonparametric regression model.Firstly,the scheduling flow of arrived flights is elaborated,and three key factors affected taxiing-time are analyzed: the capacity of taxiing-path,the designed taxiing-route and airport surface conflicts.And the mathematical simulation model of arrived flights is built,which lays the foundation of next route design and time prediction.Secondly,a part region of a large-scale hub airport is abstracted as a network topology with links and nodes,in which the key taxi nodes and gate positions are signed by different symbols.Then the taxiing-path of arrived flights is designed by ACA,the updating regulation of pheromone of ant colony algorithm is improved adaptively,which is called Adaptive ACA,and the taxiing-path is redesigned by Adaptive ACA.Combine the relevant data which is collected by intelligent algorithm and monitor date of surface,the data is get as a database of nonparametric regression model after processing,then to predict the subsequent taxi time on the nodes of the designed route according to the model of the algorithm.Finally,according to gate position allocation plan,compare the taxiing-path simulation result of intelligent algorithm to the experience scheduling,analyze the iterative process,taxi distance,taxi time and waiting time.Contrastive analysis of Mean Relative Error is made by different methods in this paper to evaluate the result of prediction time.