Bottleneck Detection on Pedestrian Flow by Modelling Spatial Dependence on Fundamental Diagrams
Recent progress in measurement methods has led to the development of fundamental diagrams (FDs) for modelling pedestrian flow.The fundamental diagrams are useful in evaluating the level of service of pedestrian facilities.However, as the velocity v depends on not only the density k but also the position x, the FD depends on the position.Therefore, there is a tendency for certain local areas to act as bottlenecks in the pedestrian facility.This study aims to detect the bottlenecks in pedestrian flow by modelling spatial dependence in FD estimation.The experiment included a mesh on the entire experimental area and using the Voronoi method to calculate local density and velocity on each mesh.Eigenvector spatial filtering (ESF) was employed to model the spatial dependence.The FD of a Greenshields equation was formulated with the ESF and the estimated parameters, which were related to both FD and spatial dependence.Meshes with smaller free-flow velocity caused by the spatial dependence term were considered as bottlenecks.The model was applied to a real flow data at a pedestrian crossing.The ESF regression resulted in lower AIC when compared with those obtained by conventional regression.The regression indicated two bottlenecks.The first bottleneck was a global bottleneck, in which one side of the crossing had a smaller capacity than the other side, even though the width of the crossing was the same.The second bottleneck was a local bottleneck, which existed at a particular corner of the crossing.
pedestrian fundamental diagram spatial dependence eigenvector spatial filtering bottleneck detection
Wataru Nakanishi Yoshiaki Fukutomi Takashi Fuse
Department of Civil Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo, 1138656, Japan
国际会议
The 8th International Conference on Pedestrian and Evacuation Dynamics (第八届行人与疏散动力学国际学术会议)
合肥
英文
277-282
2016-10-17（万方平台首次上网日期，不代表论文的发表时间）