Optimising Airport Ground Resource Allocation for Multiple Aircraft Using Machine Learning-Based Arrival Time Prediction

Managing aircraft turnaround is a complex process due to various factors, including passenger handling. Airport ground handling, resource planning, optimal manpower, and equipment utilisation are some cost-cutting strategies, particularly for airlines and ground handling service teams. Scheduled aircraft arrival and departure times are critical aspects of the entire ground management and passenger handling process. This research aimed to optimise airport ground resource allocation for multiple aircraft using machine learning-based prediction methodologies to enhance the prediction of aircraft arrival time, an uncontrollable variable. Our proposed models include a multiple linear regression (MLR) model and a multilayer perceptron (MLP)-based model, both of which are used for predicting round-trip arrival times. Additionally, we developed a MLP-based model for multiclass classification of arrival delays based on departure time and delay from the same airport. Under normal weather conditions and operational scenarios, the models were able to predict round-trip arrival times with a root mean squared error of 8 min for each origin–destination pair and classify arrival delays with an average accuracy of 93.5%. Our findings suggest that machine learning-based approaches can be used to predict round-trip arrival times based on the departure time from the same airport, and thereby accurately estimate the number of actual flight movements per hour well in advance. This predictability enables optimised ground resource planning for multiple aircraft based on constrained airport resource deployment and utilisation.