Research On Prediction Of Airport Delay Under The Influence Of Weather Based On GBRT Algorithm

With the increasing demand of air transportation,the air transportation resources are becoming more and more scarce,which leads to frequent flight delays that puzzle the relevant departments and the vast number of passengers.Using machine learning algorithm,large amounts of data generated during the flight operation are analyzed,and the airport flight delay prediction model is established,which can make the airport and other units to reduce flight delay and passenger detention in the tactical or pre-tactical stages.This paper aims to predict the delays of departure flights in the target airport,explore the factors that affect the flight delays under the influence of weather,and carry out the flight delay prediction research based on the Gradient Boosting Regression Tree(GBRT).Firstly,the Weather Impacted Traffic Index(WITI)is introduced and simplified.The Gaussian mixture clustering method is used to cluster the main weather factors that affect the flight operation.The Otsu algorithm is combined with the flight operation data to determine the main weather categories,and the weather impact traffic coefficient under each category is determined,so as to improve the traditional WITI model and make it more accurate to indicate the impact of weather on flight operations.Secondly,the airport delay spread model is established,and relevant features are extracted according to the model.The impacts of the associated airport and associated flight on the departure delays are studied from the two dimensions of time and space,and the correlation analysis is used to extract the airport flight delay spread features.Then,combining with the meteorological features and the airport flight delay spread features,adding time and flight operation features,the features of airport flight delay prediction model are determined.The GBRT is used to build the airport flight delay prediction model,and the model is analyzed and optimized by perspectives of feather data,parameter optimization and feature importance.The results show that the airport delay prediction model based on the GBRT algorithm can accurately predict the delays of airport departure flights in the next hour.Finally,the fuzzy algorithm is used to establish the corresponding membership function and fuzzy judgment rules,and the quantitative airport delay index is obtained.