对流天气下基于聚类算法的终端区交通流分析  

Analysis of Traffic Flow in Terminal Area Based on Clustering Algorithm under Convective Weather

作　　者：姚学成 胡明华[1] 袁立罡[1] 陈海燕[1] 刘振亚[1] YAO Xue cheng;HU Ming hua;YUAN Li gang;CHEN Hai yan;LIU Zhen ya(Nanjing University of Aeronautics and Astronautics,Nanjing 211000,China)

机构地区：[1]南京航空航天大学,江苏南京211000

出　　处：《航空计算技术》2022年第3期42-46,共5页Aeronautical Computing Technique

基　　金：国家重点研发计划项目资助(2021YFB1600500);国家自然科学基金项目资助(52002178)。

摘　　要：为准确分析对流天气对终端区进场交通流的影响,提出了一种面向对流天气场景的基于轨迹聚类的进场交通流分析方法。利用卷积神经网络与K means++算法对对流天气进行特征提取和聚类;采用均匀化参数方法对进场飞行轨迹进行重采样;进而采用具有噪声的基于密度的聚类算法对进场轨迹进行聚类,采用K means算法识别交通流的中心轨迹。对广州终端区历史运行数据进行实例分析,结果表明所提方法能准确识别进场轨迹与对流天气的关联性,验证了方法的有效性。In order to analyze the impact of convective weather on the arrival traffic flow in the terminal area accurately,this paper proposed an arrival traffic flow analysis method for convective weather scenes.This method is based on trajectory clustering.First,the paper uses convolutional neural network(CNN)and K means++algorithm to extract features and cluster the convective weather images.Then,uses homogenization parameter method to resample the arrival flight trajectory data to form complete trajectories.Next,density based spatial clustering of applications with noise(DBSCAN)algorithm is used to cluster the trajectories and applies K means algorithm to identifying the central trajectories of the traffic flows.Finally,the paper carries out case studies on the historical operation data of Guangzhou terminal area.The results show that the proposed method can accurately identify the correlation between the arrival flight trajectory and the convective weather.

关 键 词：终端区交通流 对流天气 卷积神经网络 轨迹聚类 

分 类 号：V355[航空宇航科学与技术—人机与环境工程]