基于CML的级联失效下异质多层交通网络节点抵抗特性研究

为了有效刻画拟合实际的多层交通网络级联失效过程，识别多层网络中对整体韧性安全影响显著的对象，提升交通网络抵抗失效的能力，建立一种基于耦合映象格子（Coupled Map Lattice， CML）的级联失效模型。首先，将交通管控措施赋予交通网络节点额外的容纳能力引申为节点抵抗特性，通过为CML添加抵抗参数H，形成考虑抵抗特性的级联失效（HCML）模型，沿常值-拓扑关联-时变拓扑关联的路径深化H的表征方式，并在小世界（Small World， SW）网络上对比3类表征方式的优劣；其次，应用Space H方法建立异质3层公共交通网络，对比度值与介数2种时变拓扑指标关联的抵抗效果，探究不同交通层抵抗系数对网络失效的影响程度；最后，在6组不同的网络结构上进行HCML模型与传统模型的对比，讨论建网方法、网络规模等不同网络性质的差异对模型有效性的影响，并结合实际客运数据验证HCML模型优越性。研究结果表明：3类表征方式中时变拓扑关联可以兼顾其余2类的优点；同等条件中网络在介数关联的抵抗特性作用下失效增速更缓，模型识别得到的对网络整体失效影响更大交通方式符合客流量实际；HCML模型在多种不同类型的多层交通网络中均可以有效实施，相比于传统模型，HCML模型的最终失效比例平均下降了5.78%，与实际交通级联失效过程更加拟合。

Node Resistance Characteristic of Heterogeneous Multi-layer Transportation Networks Under Cascading Failure Based on Coupled Map Lattice

A cascading failure model based on the coupled map lattice (CML) was developed to effectively characterize the cascade failure process of real multi-layer transportation networks, identifying the objects in the multi-layer networks that have a significant impact on the overall resilience security, and improve the ability of the transportation network to resist failure. Firstly, the resistance characteristics of the node were derived from the additional capacity given to the nodes by traffic control measures. A cascading failure model analyzing the resistance characteristic (HCML) was formed by adding the resistance parameter H to the CML, which deepened the H characterization method along the path from constant to a topological association to time-varying topological association. Moreover, the advantages and disadvantages of the three characterization methods on the small-world (SW) network were compared. Secondly, the Space H method was applied to establish a heterogeneous three-layer public transportation network and investigate the degree of influence of different transportation layer-resistance coefficients on network failure by comparing the implementation effect of two types of time-varying topological metrics, correlating degree, and betweenness. Finally, the HCML and traditional models were compared in six different groups of network structures to discuss the effects of differences in network construction methods, network scale, and other network properties on the validity of the HCML and verifying the superiority of the HCML with actual passenger transportation data. The results show that amongst the three types of characterization methods, time-varying topological correlation can consider the advantages of the remaining two types; under the same condition, the failure growth rate of the network is slower under the resistance characteristic of betweenness correlation. Additionally, the transportation mode with great influence on overall network failure obtained by the HCML identification is in line with the actual passenger flow; the HCML can be effectively implemented in different types of multi-layer transportation networks. Compared with the traditional model, the final failure ratio of the HCML decreased by 5.78% on average, making it more suitable for the actual traffic cascade failure process.