时空矢量场下人群活动聚散模式提取与分析  

Modeling human mobility and discoverying the convergence anddispersion patterns under spatial-temporal vector field

作　　者：李静[1] 刘海砚[1] 李佳[1] 陶泽坤 刘俊楠[2] 叶林 LI Jing;LIU Haiyan;LI Jia;TAO Zekun;LIU Junnan;YE Lin(Institute of Data and Target Engineering,PLA Strategic Support Force Information Engineering University,Zhengzhou 450000,China;Institute of Geoscience and Technology,Zhenzhou University,Zhengzhou 450000,China)

机构地区：[1]信息工程大学数据与目标工程学院,郑州450000 [2]郑州大学地球科学与技术学院,郑州450000

出　　处：《测绘工程》2024年第3期1-13,25,共14页Engineering of Surveying and Mapping

基　　金：国家自然科学基金资助项目(42301526);河南省自然科学基金资助项目(242300420623)。

摘　　要：由于传统方法缺乏顾及人群在区域间的动态流动而无法反映人群在未来(下一时刻)的活动聚散趋势,因此,文中借助时空矢量场来建模人群活动的趋向性,通过矢量场理论中的散度算子来定量计算人群活动聚散强度,将人群活动聚散模式提取问题转化为时间序列聚类问题识别出主要聚散模式。在海口市滴滴出行数据集上进行实验,选取角度偏态系数证明了主体方向计算方法的有效性,提取出了4种主要人群活动聚散模式,并结合POI类型的分布情况对4种模式进行了语义解释,为探索人类移动性提供研究思路和方法支持。The traditional methods usually ignore the dynamic features of human mobility,thus they could hardly reflect the trend of convergence and dispersion in the future(the next moment).In order to overcome this deficiency,spatial-temporal vector field is brought to model the dynamic features of human mobility.The aggregation and dispersion intensity of human activities is quantitatively calculated through the divergence operator in vector field,and the extraction problem of human mobility aggregation and dispersion patterns is transformed into a time series clustering problem.In this paper,experiments are carried out on Didi travel data set in Haikou City.The index(deviation coefficient)is selected to prove the effectiveness of the proposed method of main direction calculation.Four main aggregation and dispersion modes are extracted,and the semantics of the four modes are interpreted with the distribution of POIs.The proposed method is suitalble for exploring human mobility.

关 键 词：人群活动 主体方向 矢量场 聚散模式 散度 

分 类 号：P208[天文地球—地图制图学与地理信息工程]