城市多模式交通网络韧性评估研究综述

A Review about Resilience Evaluation for Urban Multimodal Transportation Networks

为促进交通韧性研究的发展,聚焦于城市多模式交通网络,对国内外韧性评估领域的相关文献进行总结。阐述了“韧性”的定义与内涵;梳理了基于网络拓扑、基于供需特性、考虑耦合关系的韧性评估指标体系;总结了模型驱动和数据驱动2类韧性评估方法的成果与优劣;探讨了网络设计、应急疏散、网络修复层面的交通网络韧性提升措施,并归纳了韧性优化的模型和算法;最后总结了现有研究不足和未来发展方向。研究结果表明:①复合网络的韧性评估未能充分考虑网络的耦合特性,韧性评估对可变的交通需求和乘客出行行为的刻画不精确;②模型驱动的韧性评估在指标权重的确定上更多依赖主观性;数据驱动的韧性评估重在数据的分析与结果展示,缺乏韧性演变规律与趋势的深度解析;③旨在提升韧性的优化模型在多目标决策、大型网络中的计算效率、真实场景的还原等方面还有待改进。未来研究的建议和展望如下:①在网络的构建、指标的获取上充分考虑复合网络的相依特性,在评估模型的构建上科学反映各系统间的耦合特性;②协同多部门建立完备共享的数据库,探索数据与模型双驱动的网络韧性评估方法,设计高效算法以支持韧性指数的快速精确计算;③将静态离散的韧性评估转化为动态连续的韧性监测,进而分析网络韧性时空演化规律与趋势,探究交通网络韧性演化机理;④精细化的网络韧性决策优化应在数据的分析和模型的构建上加强对真实事件场景的还原,并进一步探索AI智能算法在大型网络优化中的应用。

To improve the development of research about transportation resilience,this paper,focusing on urban multimodal transportation networks,summarizes the relevant studies on resilience evaluation in the literature.The definition and connotation of resilience are introduced.The indicators for resilience evaluation are summarized from the perspectives of network topology,supply-demand characteristics,and coupling relationships.The research of model-driven and data-driven resilience evaluation methods are introduced.The advantages and disadvantages of these methods are summarized as well.Fourth,measures to improve the resilience of transportation network are discussed from the perspectives of network design,emergency evacuation,and network restoration.The resilience optimization models and algorithms are summarized as well.The research deficiencies and future development directions are discussed.The results show that:①the resilience evaluation of composite networks fails to fully consider the coupling characteristics.Besides,resilience evaluation is imprecise to depict variable traffic demand and travelers'travel behavior.②The determination of indicator weights depend more on subjective judgement in model-driven resilience evaluation.Data-driven resilience evaluation focus on data analysis and result display,but lacks in-depth analysis of resilience evolution.③The optimization models targeting resilience improvement need to be improved in multi-objective decision making,computational efficiency in large-scale networks,and reproduction of real scenes.From these results,the suggestions for the future research are as follows:①in the development of the network and the selection of indicators,the dependence of the composite network needs to be fully considered.Besides,and the coupling characteristics between the systems need to be scientifically reflected in evaluation models.②It is suggested to cooperate with multiple departments to establish a complete and shared database,to explore the network resilience evaluation methods which are driven by both data and model,and to design high-efficient algorithms to support the rapid and accurate calculation of the resilience indicators.③The static discrete resilience evaluation should be developed into dynamic continuous resilience monitor,based on which the temporal-spatial evolution of network resilience and the evolution mechanism of traffic network resilience must be analyzed.④The refined network resilience decision optimization should be strengthen to reproduce the real event scenarios in data analysis and model development.Besides,it is necessary to further explore the application of AI algorithm to deal with the application of large-scale network optimization.