融合网络拓扑结构特征与客流量的城市轨道交通 关键节点识别研究

科学合理地识别轨道交通网络的关键节点，并制定针对性的维护管理措施，有助于保障轨道交通的稳定运行。网络中的节点在局部发挥着对外传输与连接的作用，在全局中影响着网络的传输效率，同时，也必然受到其他节点的影响，本文提出一个考虑邻居节点影响的改进节点度模型评价节点的局部重要性，以及一个考虑其他节点影响的改进节点效率模型评价节点的全局重要性，基于改进的节点度模型和节点效率模型构建节点网络拓扑结构重要度评价模型，综合反映节点的局部重要性和全局重要性，也能反映其他节点对目标节点的影响；以进出站客流和换乘客流为基础建立节点客流量重要度评价模型；进一步构建综合考虑节点网络拓扑结构重要度和客流量重要度的关键节点识别模型，更加全面地评价节点的重要性，并以西安市数据为基础进行实例验证。结果表明：本文模型所识别出的关键节点，能很好地体现节点在网络中的功能特性；排名前5的关键节点失效，会导致客流损失34.41%，网络效率降低57%，相对最大连通子图比例下降91.82%，证明了模型的有效性。

Identification of Key Nodes of Urban Rail Transit Integrating Network Topology Characteristics and Passenger Flow

A scientific and rational method to identify the key nodes is useful for formulating targeted management measures and the stable operation of urban rail transits. The nodes play the role of external transmission and connection locally and affect the transmission efficiency of the network globally. Since one node will inevitably be affected by other nodes in the network, an improved node degree model considering the influence of neighbor nodes was proposed to evaluate the local importance of nodes, and an improved node efficiency model considering the influence of other nodes in the network was proposed to evaluate the global importance of nodes. Based on the improved node degree model and the improved node efficiency model, a node structural importance evaluation model was constructed. This model can comprehensively reflect the local and global importance of nodes, as well as the impact of other nodes on the target node. From the perspective of passenger flow, a node flow importance model based on the inbound and outbound passenger flow and transfer passenger flow was established. Furthermore, a model considering the importance of network topology and passenger flow was conducted to identify the key nodes in urban rail transit. The proposed models were verified based on the data in Xi'an. The results show that the identified key nodes can reflect their functional characteristics in the network. The failure of the identified top 5 key nodes will result in 34.41% of passenger flow loss, 57% of network efficiency reduction, and 91.82% of the relative size of the largest connected subgraph reduction. The results indicate that the proposed models are effective and practical.