An Airport Scene Delay Prediction Method Based on LSTM

Due to the highly dynamic nature of flight operations, the prediction for flight delay has been a global problem. At the same time, existed traditional prediction models have difficulty capturing sequence information of delay, which may be caused by the subsequent transmission of delay. In this paper, a delay prediction method based on Long Short-Term Memory Model (LSTM) is proposed firstly. Furthermore, the relevant features are selected and we divide the delay levels. Then we cross-contrast performances of the model based on different hyper parameters on the actual dataset. Finally, the optimal prediction model of the scene delay is obtained. Experimental results show that compared with the traditional prediction model whose average accuracy is 70.45%, the proposed prediction model has higher prediction accuracy of 88.04%. In addition, the proposed model is verified to be robust.