Aircraft Trajectory Prediction Using Deep Long Short-Term Memory Networks

Trajectory-based operation is an air traffic control mode with more accuracy, safety, and efficiency, and an effective measure for future airspace management under the conditions of large flow, high density, and short interval, which significantly improves the utilization of airspace resources. Aircraft trajectory prediction is the key technology of trajectory-based operations, requiring the information of high-precision trajectory prediction to achieve high-density operation in airspace. The current state-of-the-art forecasting methods, which perform with low accuracy in low-altitude flight environment, are difficult to be applied in actual operation and management. In this paper, a deep long short-term memory (D-LSTM) neural network for aircraft trajectory prediction is proposed, which improves the prediction accuracy of aircraft in complex flight environments. Multi-dimensional features of aircraft trajectory are integrated into LSTM, and tested on real flight data of ADS-B, demonstrating that the proposed model has higher prediction accuracy than existing methods in different flight phases.