Interactive Temporal Convolutional Network for Spatio-temporal flight trajectory prediction and safety assessment in terminal manoeuvring area

The NextGen programme has significantly improved the reliability and predictability of air traffic management (ATM) by implementing four-dimensional (4D) trajectory prediction, which comprises spatial-temporal trajectory prediction, considering time, longitude, latitude, and altitude. This paper proposes a novel interactive methodology for real-time interactive 4D trajectory prediction to facilitate ATM. The research utilises image processing and deep learning to improve interaction capabilities with the ATM and ensure the security of the Hong Kong Flight Information Region (HKFIR). The proposed methodology includes establishing data processing and extracting holding patterns utilising image processing. The proposed Interactive Temporal Convolutional Network (ITCN) is compared with several popular time- series prediction models, including the Recurrent Neural Network (RNN), the Long Short- Term Memory (LSTM) neural network, the Gated Recurrent Unit (GRU) neural network, and the Transformer for short-term spatio-temporal flight trajectory prediction.