基于YOLO的人工智能飞机尾涡识别研究

Research on Aircraft Wake Vortex Recognition Based on YOLO Artificial Intelligence

为有效保障空中交通安全,提高机场跑道利用率,设计一种基于人工智能的算法用于识别近地阶段的飞机尾涡,将识别结果可视化后提供给空中交通管制员,辅助管制员作出指令决策,为缩减现行雷达管制间隔标准提供依据。该方法采用单步目标检测,直接通过网络产生目标位置和类别信息,将机场区域采集到的尾涡数据输入模型进行识别。结果表明:提出的人工智能算法能够有效识别飞机进近着陆阶段的飞机尾涡,并且结合特征金字塔的网络结构实现了预测值与真实值的高度相关性,显著提高飞机尾涡目标检测精度,能够为空中交通管制员提供高准确度的机场区域内尾涡演化情况,在安全的情景下,结合实际大气特性为缩短现行的雷达管制间隔提供支撑。

In order to keep air traffic safety and improve airport runway utilization,an artificial intelligence-based algorithm was designed to identify the aircraft wake vortex in the near-field phase.After giving visualized results to air traffic controller,the assistant controller can made decisions providing a basis for reducing existing radar control interval standards.The method adopts the single-step target detection law to directly generate the position and category information of the target through the network,and input the wake vortex data collected at the airport into the model for identification.The results show that the artificial intelligence algorithm proposed in this paper can capably identify the aircraft wake vortex in the near-field phase,and combines the network structure of the Feature Pyramid Networks to achieve a high correlation between the predicted and the real value.Compared with the traditional way,the target detection can be improved.It provides air traffic controllers with high accuracy of wake vortex evolution in the airport area and,in a safe condition,combined with actual atmospheric characteristics to support the shortening of current radar control intervals.