一种基于多模型融合的飞机尾涡预测方法
By combining multi-model fusion and deep learning techniques with traditional physical models and neural networks, the short-term and long-term characteristics of wake vortices are identified, solving the problems of low accuracy and efficiency in wake vortex prediction and achieving reliable prediction of dynamic wake intervals.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CIVIL AVIATION UNIV OF CHINA
- Filing Date
- 2026-04-28
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies for predicting wake vortices suffer from low accuracy and slow computational efficiency. In particular, in the context of flight takeoffs and landings at large, busy airports, it is difficult to achieve dynamic wake vortex spacing, which affects aircraft operating efficiency and airport capacity.
A multi-model fusion approach is adopted, combining lidar data and various traditional physical models. Convolutional neural networks (CNN) and bidirectional long short-term memory neural networks (BiLSTM) are used to identify the short-term and long-term characteristics of the wake vortex. Bayesian optimization and Monte Carlo dropout techniques are used to improve the prediction accuracy and interpretability.
It improves the accuracy and efficiency of wake vortex prediction, provides an interpretable safety margin range in airport environments with high real-time requirements, and supports the implementation of dynamic wake separation.
Smart Images
Figure CN122242273B_ABST