基于深度概率模型的通用机场航空器轨迹预测OA

Aircraft trajectory prediction is a key technology for airports to ensure low-altitude flight safety and improve operational efficiency.General aviation flights are characterized by diverse tasks,flexible paths,and environmental sensitivity.Traditional prediction methods struggle to accurately model these complex characteristics.This paper proposes a trajectory prediction framework based on deep probability models,and employs conditional variational inference methods to model the probability distribution of trajectories.By introducing latent variables to capture the potential factors influencing flight trajectories and integrating physical constraints and trajectory smoothing regularization,the framework ensures the generated trajectories conform to the aircraft's dynamic characteristics and flight rules.Experimental results show the proposed method reduces the displacement error index by 18.3%and 16.7%respectively compared other advanced benchmark methods.The ablation experiments validate the effectiveness of the modules.Among them,variational inference contributes the most(improving performance by 15.2%)and the environmental perception module also plays a critical role(improving performance by 8.7%),ensuring the safe operation of airports.