Research On The Prediction Of Taxi-out Time Of Departures From Dual Runway Airports

The problem of airport scene congestion has become the main factor restricting air traffic and transportation.With the rapid increase of flight flow,the average taxiing time of scene also increases.At present,the taxiing time of large airports has exceeded 25 minutes,and taxi conflicts often occur,which seriously affects the operation efficiency of airport scene.Therefore,it is particularly important to predict the slide-out time of flights scientifically and accurately,improve the operation efficiency and coordination decision-making ability of the airport scene,optimize the take-off sequence of flights,improve the traffic condition of the airport scene,reduce the operation cost,and accelerate the traffic flow of the scene.This paper starts with the operation process of airport and expounds the importance of accurate prediction of departure time.Various factors affecting the departure time of the flight were analyzed.After introducing related concepts in A-CDM system,8 quantifiable factors were selected for definition.Then,the actual operation data is filtered,calculated,analyzed and quantified.k-means cluster analysis was used to analyze the gliding efficiency of the processed data,and SPSS and Minitab statistical analysis software were used to test the correlation between eight quantifiable influencing factors and the actual sliding time.Based on the results of correlation analysis,the BP neural network and support vector machine(SVM)slip out time prediction model were constructed,and the advantages and disadvantages of the two models were analyzed,and the evaluation criteria were established.Based on the data of an airport in Central South,the experimental results show that:(1)The number of launching aircraft in the same period,the number of taking off aircraft in the same period,the number of landing aircraft in the same period,the average slide-out time of 1hour and the slide-out time of departing aircraft show strong correlation.The slide-out time is correlated but not significant with the slide-out distance,the number of turns and delay time.The time period of the aircraft is not correlated with the slide-out time.(2)The 1-hour average slip out time plays an important role in the improvement of model accuracy.(3)The trend of prediction results based on SVM and BP neural network is consistent,and the accuracy of seven-tuple prediction results considering strong correlation and correlation but not significant correlation factors is the best.If weak or irrelevant factors are added,the accuracy of octuple prediction will decrease significantly.(4)The accuracy of slip-out time prediction results based on SVM is significantly higher than that based on BP neural network.The mean absolute error percentage(MAPE)is only 0.1186,the mean absolute error(MAE)is 92.2674 s,and the mean square error(RMSE)is 120.5878 s.The accuracy of slip out time within ±3min is 89.5%,and the accuracy of slip out time within ±5min is 98%.(5)The calculation time of the slideout time prediction model based on SVM is obviously higher than that based on BP neural network,and the time complexity is higher.