Deep Learning-Based Flight Trajectory Prediction Using Time Series Decomposition

Flight trajectory prediction of unmanned aerial vehicles (UAVs) is essential for air traffic control. Automatic dependent surveillance-broadcast (ADS-B) has been widely used in civil aviation aircraft. However, effective trajectory prediction mechanisms and sufficient data sources for ADS-B information are still lacking specifically for UAVs. This paper proposes a deep learning-based flight trajectory prediction framework using the neural hierarchical interpolation for time series (N-HITS). This framework includes ADS-B flight trajectory data generation, flight trajectory data processing, and N-HITS prediction neural network. Comparative results with existing methods like Gaussian process regression (GPR) and long short-term memory (LSTM) network indicate that the proposed framework has better generalization ability, prediction accuracy, and stability.