Search

Abstract In this work, a machine translation framework is proposed to tackle the flight plan generation in the air transport field. Diverging from the traditional human expert‐based way, a novel sequence‐to‐sequence transfer transformer network to automatic flight plan generation with enhanced operational acceptability is presented. It allows the user to translate the departure and arrival airport pairs denoted as test sentences, into the flyable waypoint sequences denoted as the corresponding source sentences. The approach leverages deep neural networks to autonomously learn air transport specialized knowledge and human expert insights from industry legacy data. Moreover, a multi‐head attention mechanism is adopted to model the complex correlation between airport pairs. Besides, we introduce an innovative waypoint embedding layer to learn effective embeddings for waypoint sequences. Additionally, an extensive flight plan dataset is constructed utilizing real‐world data in China spanning from July to September 2019. Employing the proposed model, rigorous training and testing procedures are conducted on this dataset, yielding remarkably favourable outcomes based on automatic evaluation metrics that are BLEU and METEOR, which outperform other popular approaches. More importantly, the proposed approach achieves high performance in the operational validation and visualization, showing its application potential for real‐world air traffic operation.

Accurate air traffic flow prediction assists controllers formulate control strategies in advance and alleviate air traffic congestion, which is important to flight safety. While existing works have made significant efforts in exploring the high dynamics and heterogeneous interactions of historical air traffic flow, two key challenges still remain. (1) The transfer patterns of air traffic are intricate, subject to numerous constraints and limitations such as controllers, flight regulations, and other regulatory factors. Relying solely on mining historical traffic evolution patterns makes it difficult to accurately predict the constrained air traffic flow. (2) Weather conditions exert a substantial influence on air traffic, making it exceptionally difficult to simulate the impact of external factors (such as thunderstorms) on the evolution of air traffic flow patterns. To address these two challenges, we propose a Spatiotemporal Knowledge Distillation Network (ST-KDN) for air traffic flow prediction. Firstly, recognizing the inherent future insights embedded within flight plans, we develop a “teacher-student” distillation model. This model leverages the prior knowledge of upstream-downstream migration patterns and future air traffic trends inherent in flight plans. Subsequently, to model the influence of external factors and predict air traffic flow disturbed by thunderstorm weather, we propose a student network based on the “parallel-fusion” structure. Finally, employing a feature-based knowledge distillation approach to integrate prior knowledge from flight plans and extract meteorological features, our method can accurately capture complex and constrained spatiotemporal dependencies in air traffic and explicitly model the impact of weather on air traffic flow. Experimental results on real-world flight data demonstrate that our method can achieve better prediction performance than other state-of-the-art comparison methods, and the advantages of the proposed method are particularly prominent in modeling the complicated transfer pattern of air traffic and inferring nonrecurrent flow patterns.