基于NCut算法的道路网子区划分研究  

Road Network Subarea Division Based on NCut Algorithm

作　　者：李明东 秦子雁 LI Mingdong;QIN Ziyan(Aisino Corporation,Beijing 100195,China;School of Traffic and Transportation,Beijing Jiaotong University,Beijing 100044,China)

机构地区：[1]航天信息股份有限公司,北京100195 [2]北京交通大学交通运输学院,北京100044

出　　处：《交通节能与环保》2020年第6期46-48,共3页Transport Energy Conservation & Environmental Protection

摘　　要：在对区域路网进行宏观研究时,由于交通流的复杂性和随机性,容易造成路网密度分布不均。因此,本文基于MFD理论,在利用MFD理论进行边界控制之前,提出基于归一化割(Normalized Cut,NCut)的路网子区划分算法将"异质"的路网划分为多个具有"同质性"的子区。首先,基于交通运行相似度构建了路网带权图,分析了NCut算法流程。其次按照一定的原则,采用NCut算法对以组间相似度为标准对路网进行近似归一划割。接着,为了克服NCut算法每次无法判断划分子区数量的问题,本文引入了各子区间匀质度的均值NSk(A,B)为评价指标来确定划分的子区数。最后,实验分析表明,根据该模型对望京区域路网进行划分,发现划分为4个子区时NS值最优。Due to the complexity and randomness of traffic flow,it is easy to cause uneven distribution of road network density in the macroscopic study of regional road network.Therefore,based on THE MFD theory,prior to the boundary control based on THE MFD theory,the Normalized Cut(NCut)network sub-district classification algorithm divides the"heterogeneous"networks into multiple"homogeneous"sub-districts.Firstly,based on the similarity of traffic operation,a road network weighted map is constructed and the flow of NCut algorithm is analyzed.Secondly,according to certain principles,NCut algorithm is used to conduct approximate normalized cutting of road network based on the similarity between groups.In order to overcome the problem that NCut algorithm cannot judge the number of delimiting molecular areas each time,this paper introduces the mean value of the evenness of each sub-interval as the evaluation index to determine the number of subregions.Finally,an experimental analysis is carried out,and the wangjing regional road network is divided according to the model.It is found that NS value is optimal when divided into four sub-regions.

关 键 词：Ncut算法 交通运行相似度 路网带权图 匀质度均值 

分 类 号：U491.13[交通运输工程—交通运输规划与管理]