Research On Flight Delay Prediction In Terminal Area Based On Multidimensional Data

Flight delays has become one of the key issues of concern in civil aviation industry.The terminal area,as a node in the civil aviation transportation network,is the main place where flight delays occur.Effectively predicting flight delays in terminal areas,especially departing flight delays,has positive practical value for alleviating flight delays,improving service level of civil aviation flights and improving passengers’ satisfaction with civil aviation.In this paper,the departure flight in the terminal area is used as the research object.The collected flight delays related data are used to analyze the delay characteristics of the departure flight in the terminal area by method of statistical analysis.A prediction model of flight delay in the terminal area based on deep neural network is designed.It provides theoretical and methodological support for the delay analysis and delay prediction of departure flights in the terminal area.The main research contents are as follows:Firstly,through sorting and analyzing the research of domestic and foreign researchers on flight delay and prediction,it is found that domestic and foreign researchers mainly use data mining methods such as machine learning and deep learning to mine flight data and meteorological data to achieve flight delay prediction.Secondly,compare and analyze the definitions of flight delays by civil aviation authorities at home and abroad,analyze the main factors affecting domestic flight delays,and get the conclusion that meteorological factors,airline factors and air traffic control factors are the main factors that affect flight delays through data statistical analysis.Then summarize the general process of flight delays prediction.Thirdly,flight data,meteorological data,corridor data and flight segment data are selected for research on flight delays characteristics analysis and flight delays prediction.A data fusion algorithm is designed and implemented based on the temporal and spatial relationship between flight data and meteorological data.Fourthly,by analyzing the characteristics of departure flight delays,it is found that the departure flight delays concentrated below 90 minutes,and its overall conforms to the exponential distribution.It is found that there is a strong correlation between the characteristics of departure flight delays in terminal area and that in airlines and departure directions.Comparation and analysis between the delay rate and the average departure delay time of departure flights under severe weather conditions and the overall delay rate and the overall average delay time,has found that the departure flight delays rate is relatively large and average delay time is more longer when the weather conditions are relatively bad,which indicates that the weather conditions have an important impact on the departure flight delays.Finally,a prediction model of departure flight delays based on deep neural network is proposed.Comparing the prediction effects of models with different hidden layer units,the results of simulation experiments show that with the increase of the number of hidden layer units,the prediction effect of the model has not significantly improved,but the time complexity of the model has continued to increase.Comparing the prediction results of the model when inputting different data,the results show that with the addition of the flight sequence number and departure data,the prediction accuracy of the model and the prediction accuracy at various delay levels can be improved to a certain extent.The final prediction accuracy of model can reach 89.9%,indicating that the model is feasible and has higher practical value.