摘要
随着航空事业的发展,对航迹进行聚类分析,存在许多应用价值。在分析历史飞行航迹特征的基础上,将航迹看作时间序列,采用近邻传播聚类算法,对航迹进行聚类分析,得到聚类结果并进行优化分析。近邻传播算法(AP)是建立在相似度矩阵基础上进行的聚类,为了得到相似度矩阵,结合航迹不等长的特征,选择使用DTW距离作为航迹间相似性的度量;同时,使用DCT对航迹时序列进行降噪,以求得到更好的聚类效果。实验结果表明:该方法在393条航迹的数据集中,划分出11个聚类,提高了航迹聚类的准确性。
With the development of aviation industry,clustering analysis of trajectories has much application value.Based on the analysis of the characteristics of historical flight trajectories,this paper regards the trajectory as time series,and uses AP(Affinity Propagation)clustering algorithm to cluster the trajectories and get clustered results for optimization analysis.The AP algorithm is based on the similarity matrix.In order to get the similarity matrix,the DTW(Dynamic time warping)distance is selected as the measure of the similarity between trajectories,for which are of unequal length.Besides,we use DCT(Discrete Cosine Transform)to denoise the track time series to get a better clustering effect.The experimental results show that the method is divided into 11 clusters in the data set of 393 tracks,which improves the accuracy of track clustering.
作者
张榆薪
王欣
ZHANG Yu-xin;WANG Xin(School of Computer Science,Civil Aviation Flight University of China,Guanghan 618307,China)
出处
《软件导刊》
2018年第4期89-90,100,共3页
Software Guide
基金
中国民用航空飞行学院科研基金面上项目(J2016-31)
中国民用航空飞行学院2016年度学院学生基金项目(X2016-70)
