Prediction of Arrival Flight Operation Strategies under Convective Weather Based on Trajectory Clustering

Abstract

An airport’s terminal area is the bottleneck of the air transport system. Convective weather can seriously affect the normal flight status of arrival and departure flights. At present, pilots take different flight operation strategies to avoid convective weather based on onboard radar, visual information, adverse weather experience, etc. This paper studies trajectory clustering based on the OPTICS algorithm to obtain the arrival of typical flight routes in the terminal area. Based on weather information of the planned typical flight route and flight plan information, Random Forest (RF), K-nearest Neighbor KNN (KNN), and Support Vector Machines (SVM) algorithms were used for training and establishing the Arrival Flight Operation Strategy Prediction Model (AFOSPM). In this paper, case studies of historical arrival flights in the Guangzhou (ZGGG) and Wuhan (ZHHH) terminal area were carried out. The results show that trajectory clustering results based on the OPTICS algorithm can more accurately reflect the regular flight routes of arrival flights in a terminal area. Compared to KNN and SVM, the prediction accuracy of AFOSPM based on RF is better, reaching more than 88%. On this basis, six features—including 90% VIL, weather coverage, weather duration, planned route, max VIL, and planned Arrival Gate (AF)—were used as the input features for AFOSPM, which can effectively predict various arrival flight operation strategies. For the most frequently used arrival flight operation strategies under convective weather conditions—radar guidance, AF changing, and diversion strategy—the prediction accuracy of the ZGGG and ZHHH terminal areas can exceed 95%, 85%, and 80%, respectively.

Terminal Traffic Situation Prediction Model under the Influence of Weather Based on Deep Learning Approaches

“…Appeal research has carried out an effective prediction on the delay level under the influence of weather by constructing a classification prediction model. For deep learning methods prediction, Klein et al [19][20][21] proposed a new prediction model based on the weather-affected traffic index. They first quantified the impact of convective weather on airport operations, and then used delay features and deep learning methods to predict airport delays.…”