Flight Off-Block Time Prediction Based On Cascaded Model

A reasonable pre-departure sequencing concept element of flights can improve the efficiency and predictability of airport,airline and air traffic control,and reduce the waiting time before departure.And,airlines express the preference of flight departure through the Target Off-Block Time of flight.The TOBT(Target Off-Block Time)is a point in time to be monitored and confirmed by the airline/handling agent at which all aircraft doors are closed,all passenger boarding bridges have been removed from the aircraft and thus start-up approval and push-back/taxi clearance can be received.The timely,accurate and stable TOBT of flight is a prerequisite for establishing pre-departure sequencing concept element,which has important decision significance for adjusting flight departure order and calculating the departure time of flight.By studying the process of ground handling,the flight off-block time is predicted at different times of the ground handling process,which provides a reference for estimating Target Off-Block Time of flight.Through the research,it is found that there is a certain correlation between the flight off-block time,the milestone approach,and the ground handling.Therefore,a prediction model based on cascaded BP Neural Network is proposed to predict the off-block time of the flight.First,the off-block time of flight is predicted at different times during the process of ground handling by the cascade model which uses BP Neural Network as the model component.Then,the study of the turn-around time on different flights show that the problem of overfitting can be solved by data partition.The experimental results show that the prediction accuracy of the proposed model within the tolerance of 15 minutes can reach 84.9% before the time of flight arrival,and reach 95% or more at the time of light entry,the flight check-in and the cabin door close.In the course of the research,it is found that the departure time of the flight has a certain influence on the prediction accuracy of the flight off-block time.Meanwhile,Initializing the weights and thresholds of BP Neural Networks by random numbers often leads to the instability of the prediction results and makes the final results not optimal.To solve the problem,the influence of flight departure time on the prediction accuracy of flight off-block time show that the experimental data can be divided according to the different departure time.And then,Genetic Algorithm is used to optimize the weight and threshold of BP Neural Network.The experimental results show that the BP Neural Network optimized by Genetic Algorithm not only makes the prediction result more stable,but also has higher prediction precision.