Research On Civil Aviation Flight Delay Prediction Method Based On Recurrent Neural Network

With the continuous development of economy,people’s demand for quality of life is getting higher and higher.Civil aviation transportation has become the first choice for long-distance travel by virtue of its comfort and speed.However,the rapid growth in the number of passengers and the limited airline resources sometimes cannot be matched with each other,resulting in flight delays from time to time.To this end,this dissertation proposes a recurrent neural networkbased flight delay prediction model with the goal of flight delay prediction and deep learning techniques.The model predicts flight delay status based on flight operation data and weather data,and provides a reference for relevant departments.The main studies are as follows:(1)Based on the flight approach and departure operation flow,we analyze the main factors causing flight delays and the propagation of flight delays,and design a pre-processing process for flight operation data and weather data.The process uses Spearman correlation analysis to analyze the correlation of flight delay factors and screen out the important features with high correlation with flight delays.Based on this,the time series data set construction is completed by fusing flight operation data and weather data.(2)A flight delay prediction model based on long-and short-term memory neural network is constructed for the characteristics of flight delay dataset with strong temporal order and combined with attention mechanism,and the hyperparameters of the model are optimized by comparison experiments.The correlation analysis shows that the long and short-term memory neural network has higher accuracy compared with other neural network models in the flight delay prediction problem.(3)To further improve the model prediction accuracy,a flight delay prediction model based on temporal convolutional network and long and short-term memory neural network is constructed.The temporal convolutional network is used to extract features from meteorological data with strong temporal sequences and many feature dimensions.Combined with flight operation data,flight delays are predicted more effectively.After experiments to select the optimal model parameters,the model is compared and analyzed with different neural network models,and the model has higher prediction accuracy.The flight delay prediction model based on temporal convolutional network and long and short-term memory neural network constructed in this dissertation can predict the flight delay status more accurately based on historical flight operation data and weather data,and provide scientific decision support for relevant aviation departments to reduce the adverse effects of flight delays.