Identification of Critical Nodes in Urban Transportation Network Through Network Topology and Server Routes

The identification of critical nodes has great practical significance to the urban transportation network (UTN) due to its contribution to enhancing the efficient operation of UTN. Several existing studies have discovered the critical nodes from the perspectives of network topology or passenger flow. However, little attention has been paid to the perspective of service routes in the identification of critical stations, which reflects the closeness of the connection between stations. In order to address the above problem, we propose a two-layer network of UTN to characterize the effects of server routes and present a novel method of critical nodes identification (BMRank). BMRank is inspired by eigenvector centrality, which focuses on network topology and mutual enhancement relationship between stations and server routes, simultaneously. The extensive experiments on the UTN of Shanghai illustrate that BMRank performs better in the identification of critical stations compared with baseline methods. Specifically, the performance of BMRank increases by 12.4% over the best of baseline methods on a low initial failure scale.