基于动态运行特征的空中管制扇区分类研究

Classification of Air Traffic Control Sectors Based on Dynamic Operation Characteristics

对管制扇区进行科学分类,有助于了解不同类别扇区的运行特征,为管制运行管理提供优化依据。针对国内外在扇区分类方面研究的匮乏和分类专项管理较为粗放的现状,通过模糊聚类和遗传算法等机器学习算法,研究了一种基于动态运行特征的空中管制扇区分类方法。基于ADS-B雷达航迹数据,建立反映扇区动态运行特性的指标集。引入核函数和遗传算法改进模糊C均值聚类算法,运用改进的算法对选取的中国华东和中南地区97个扇区进行分类研究。采用Silhouette系数和KXB(Kernel Xie-Beni)指标确定最佳聚类数,根据聚类结果定义出5类运行特征各异的扇区,并通过仿真实验对扇区分类成果进行了合理性检验。研究结果表明:改进的算法在IRIS和WINE数据集上的聚类准确度较传统FCM算法提高了4.62%和26.96%,且具有较高的鲁棒性;扇区动态运行特征分析是将航空器空间行为和扇区繁杂程度及管制员工作负荷进行关联比较的有效途径;不同类型扇区的运行特征存在明显差异,模糊聚类算法相比于传统的硬聚类算法能有效处理扇区分类问题。

The scientific classification of control sectors is helpful to understand the operation characteristics of different types of sectors and provide optimization basis for regulation and operation management.In view of the lack of research on sector classification at home and abroad and the current situation of extensive classification special management,an air traffic control sector classification method based on dynamic operation features was studied through machine learning algorithms such as fuzzy clustering and genetic algorithm.Based on the ADS-B radar track data,the index set reflecting the dynamic operation characteristics of the sector was established.Kernel function and genetic algorithm were introduced to improve the fuzzy C-means clustering algorithm,and the improved algorithm was used to classify 97 sectors in East China and Central South China.The optimal clustering number was determined by Silhouette coefficient and KXB(Kernel Xie-Beni)index,and five types of sectors with different operating characteristics were defined according to the clustering results.The rationality of sector classification results was tested by simulation experiments.The research results show that:the clustering accuracy of the improved algorithm on IRIS and WINE data sets is 4.62%and 26.96%higher than that of the traditional FCM algorithm and the improved algorithm has high robustness.The dynamic operation characteristics of the sector is an effective way to correlate and compare the spatial behavior of the aircraft,the complexity of the sector and the workload of the controller.There are obvious differences in the operation characteristics of different types of sectors,which shows that the fuzzy clustering algorithm is effective in dealing with the practical problem of sector classification compared with the traditional hard clustering algorithm.