Cascading failure analysis and robustness optimization of metro networks based on coupled map lattices: a case study of Nanjing, China

Document Type

Journal Article

Publication Date

2021

Subject Area

place - asia, place - urban, mode - subway/metro, planning - safety/accidents, planning - methods

Keywords

Metro networks, Cascading failures, Robustness optimization, Coupled map lattices, Passenger flow management

Abstract

Cascading failure in metro networks is a dynamic chain process induced by the interaction of passenger flow and network topology. In this paper, a bi-directional coupled map lattice model is proposed to study the cascading failure of metro networks. The model considers the two-way traffic problem, and the results are closer to those of actual metro networks than the previous one-way coupling models. A η-based flow redistribution method is proposed, and different passenger flow redistribution strategies after station failure can be achieved by changing the flow redistribution coefficient η from 0 to 1. Moreover, the robustness of metro networks can be optimized by searching for the optimal η that can maximize the critical perturbation leading to global network failure. We study the actual case of Nanjing metro. The analysis results show that the network is more vulnerable to intentional attacks than to random failures, and global network failure is triggered more easily on the largest strength station than on the stations with the largest betweenness and largest degree. The influence of coupling strengths on the critical perturbation is also investigated. The results show that larger coupling strengths correspond to smaller critical perturbations, but a change in the coupling strengths has a small impact on the optimal η. Under the given traffic data, the optimal η for Nanjing metro is approximately in the range (0.3, 0.4). This study provides a reference for developing strategies for dynamic safety evaluation and emergency management of passenger flow in metro networks.

Rights

Permission to publish the abstract has been given by SpringerLink, copyright remains with them.

Share

COinS