基于复杂网络的高速公路重要路段识别模型  

Expressway Important Section Identification Model Based on Complex Network

作　　者：温志勇 翁小雄[1] 张鹏飞 WEN Zhi-yong;WENG Xiao-xiong;ZHANG Peng-fei(School of Civil Engineering&Transportation,South China University of Technology,Guangzhou,Guangdong 510641,)

机构地区：[1]华南理工大学土木与交通学院,广东广州510641

出　　处：《公路交通科技》2024年第12期1-8,76,共9页Journal of Highway and Transportation Research and Development

基　　金：广东省重点领域研发计划项目(2022B0101070001)。

摘　　要：随着高速公路道路结构的日益复杂,识别高速公路重要路段对于交通特性研究及运营管理至关重要。基于多层复杂网络原理从交通系统的供给和需求两个方面出发,构建高速公路重要路段识别模型。模型由多层复杂网络、路段重要度评估指标以及基于熵权法的K-means聚类等3个模块组成。首先,针对高速公路交通系统的复杂特性,建立由结构网、权重网和出行网组成的多层复杂网络,以准确反映路段的道路结构、通行成本和交通出行等多种属性,实现交通系统精细化建模;其次,在多层复杂网络基础上构建表征路段交通供需特征的路段重要度指标,具体地,在权重网上采用K条最短路径形成有效路径集合结合多项Logit算法进行路径选择,从而构建路段概率介数,以准确反映存在多义路径情况下的路段在交通供给侧的重要性;在出行网上构建路段流量介数,以反映路段在交通需求侧的重要性,进一步结合路段权重的自身特性构建加权概率介数与加权流量介数,形成路段重要度的多属性指标;然后,针对路段的多属性指标研究熵权法权重聚类算法,以聚类方式实现高速公路重要路段分类,避免等量划分的主观性。最后,以粤港澳大湾区高速公路为研究对象进行案例分析,模型结果经过SIR网络传播模型和对比分析验证了模型的有效性和可靠性。As expressway structures become increasingly complex,identifying the expressway important section is essential for traffic characteristic analysis and operational management.Focusing on both the supply and demand aspects of transportation system,the expressway important section identification model was developed based on the principle of multi-layer complex network.The model consists of 3 modules:multilayer complex network,section importance assessment metrics,and entropy weighted K-means clustering.First,in view of the complexity of expressway transportation system,the multilayer complex network was established,comprising the structural network,weighted network,and travel network.The multilayer network effectively captured the structural,cost-related,and travel attributes of each section,realizing a more precise model of transportation system.Second,the importance metrics representing the supply and demand characteristics of each section were developed based on this multilayer complex network.On the weighted network,a set of effective paths was generated by using the K-shortest paths.Combining with the multinomial Logit model for route choice,the section probability betweenness was constructed to accurately reflect a section’s significance on transportation supply under conditions of multiple route choices.On the travel network,the section flow betweenness was created to capture its importance on transportation demand.Further,constructed weighted probability betweenness and weighted traffic betweenness by combining the inherent characteristics of section weights,forming the multi-attribute importance metrics for each section.Then,the entropy weighted clustering algorithm was studied,implementing the classification on expressway important sections by using clustering,and reducing the subjectivity of equal-interval division.Finally,the case study of expressways in the Guangdong-Hong Kong-Macao Greater Bay Area demonstrated the model’s effectiveness and reliability.The result was validated through the SIR netwo

关 键 词：智能交通 重要路段识别模型 复杂网络 路段重要性 熵权法聚类 加权介数 

分 类 号：U495[交通运输工程—交通运输规划与管理]