Research On Airline Flight Delay Prediction And Recovery Strategies

Flight delays are a major concern in the civil aviation industry.In practice,flight delays can lead to serious losses to airlines and passengers’ interests.In order to reduce the economic losses arising from flight delays,this paper investigates the problem of delay prediction and recovery of departing flights based on historical flight data and airport meteorological data from Chengdu Shuangliu Airport.The details of the study are as follows:Firstly,the raw data of historical flight data and airport meteorology of Chengdu Shuangliu Airport were processed,and the factors influencing flight delays were analyzed on the basis of this data,focusing on the impact of bad weather conditions on flight delays and the ranking of meteorological factors affecting flight departures.Data fusion by using Python and Excel for data;According to the time-series characteristics of operational big data,the LSTM network and GRU network are selected to build the prediction models according to the good processing ability of recurrent neural network for time-series data;meanwhile,the attention mechanism is added to enhance the prediction effect of the neural network model.In this paper,four departure flight delay prediction models,namely,LSTM model,GRU model,Attention-LSTM model and Attention-GRU model,are constructed;by comparing the training iteration speed of the above four models,it is verified that GRU has the fastest processing speed for aviation operation big data;by comparing the performance evaluation indexes of the above four models,it is verified that the neural network model with By comparing the performance indicators of the above four models,it is verified that the training error of the neural network model with the attention mechanism is smaller and the prediction accuracy is significantly improved;and by comparing the indicators,it is verified that the Attention-GRU model is the best prediction model in the practical application.Finally,a minimum delay loss cost recovery model was developed and solved using a dynamic programming algorithm.Through case validation,it is demonstrated that the flight recovery strategy proposed on the basis of the model has a significant effect on the reduction of flight delay loss costs.