A Deep Graph-Embedded LSTM Neural Network Approach for Airport Delay Prediction

Due to the strong propagation causality of delays between airports, this paper proposes a delay prediction model based on deep graph neural network study from perspective an airport network. We regard airports as nodes and use directed construct airports’ relationship. For adjacent weights edges are measured by spherical distance them, while number flight pairs them is utilized for connected flights. On basis, diffusion convolution kernel constructed capture characteristics it further integrated into sequence-to-sequence LSTM establish learning framework prediction. name graph-embedded (DGLSTM). To verify model’s effectiveness superiority, we utilize historical data 325 in United States 2015 2018 training set test set. The experimental results suggest that proposed method superior existing mainstream methods terms accuracy robustness.