Bottleneck Identification Method of Urban Rail Transit Network Based on Spectral Clustering

Ruoyu Zhao,Jun Liu

DOI: https://doi.org/10.1109/icectt.2019.00084

2019-01-01

Abstract:The capacity bottleneck of urban rail transit network restricted the improvement of the capacity of the road network. Identifying capacity bottlenecks efficiently plays an important role in the new line planning of the rail transit planning department and adjustment of capacity resources of the operation department. Based on spectral clustering, this paper proposed a method to identify the bottleneck of urban rail transit, which can quickly and accurately identify the critical bottleneck of the urban rail transit network. Firstly, the transfer network was constructed with the transfer stations as the nodes to effectively reduce the computational complexity. Furthermore, in order to satisfy the input requirements of spectral clustering, a transfer network construction method based on undirected graph was proposed. Then, the network topology and section capability were selected as clustering indicators. To solve the problem, this paper proposed a model which was established based on normalized-cut criterion. After a case study on a city's rail transit network, the result shows that under the changed of capacity, the spectral clustering algorithm could be used to identify the capacity bottleneck of urban rail transit network and the result of identification can accurately reflect the critical bottleneck of the network. Compared with the traditional method of scanning network, the calculation efficiency and recognition effect of this method are greatly improved.