面向多元时空对象聚类分析的道路工程统筹

Integrated Road Engineering Planning for Multiple Spatio-temporal Objects Clustering Analysis

传统时空聚类分析主要面向轨迹数据和点状事件,在实际应用中具有局限性。提出了一种基于ST-DBSCAN的多元时空对象聚类分析方法,针对点、线、面等多元化时空对象的空间形态和存续时间,构建度量对象相似性的时空距离模型,并提供关键参数——时空邻域半径的确定策略。在时空距离模型方面,分别计算时空对象的时间距离和空间距离,从而建立综合性的时空距离;在参数确定方面,综合了K-dist图和曲线拐点求取方法,使得聚类结果更加合理和准确。针对城市道路工程的应用结果表明,该方法能获得具有很高内部时空相似性的聚类簇,可有效指导道路工程统筹协调。

The traditional spatio-temporal clustering analysis mainly focuses on trajectory data and point events,which is limited in practical application.In this paper,we proposed a clustering analysis method for multiple spatio-temporal objects based on ST-DBSCAN.We constructed a novel spatio-temporal distance model to measure the similarity of spatio-temporal objects according to the spatial morphology and existence time of multiple spatio-temporal objects such as point,line and surface,and provided the determination strategy of spatio-temporal neighborhood radius.In the aspect of spatio-temporal distance model,we calculated the temporal distance and spatial distance of spatio-temporal object respectively,so as to establish the comprehensive spatio-temporal distance.In terms of parameter determination,we integrated K-dist figure and curve inflection point calculating method to make the clustering results more reasonable and accurate.The application results of urban road engineering show that the spatio-temporal clustering method can obtain clustering clusters with high internal spatio-temporal similarity,and effectively integrate all kinds of road engineering.