铁道运输与经济2025，Vol.47Issue(5)：96-107,12.DOI:10.16668/j.cnki.issn.1003-1421.2025.05.09
基于图神经网络节点强化的四川省综合立体交通网络优化研究
Optimization for Comprehensive Three-Dimensional Transportation Network in Sichuan Province Based on Node-Enhanced Graph Neural Network
摘要
Abstract
The Chengdu-Chongqing economic circle,one of the"four poles"in China's comprehensive three-dimensional transportation network backbone,has experienced progressive development yet faced challenges like insufficient network connectivity and poor robustness,necessitating network optimization.By using complex network theory,the construction and topological feature analysis of the comprehensive three-dimensional transportation network were conducted.A node reinforcement optimization method based on graph neural networks was developed,and simulation analysis was carried out by taking the comprehensive three-dimensional transportation network in Sichuan Province as an example.The results indicate a 20%improvement in closeness centrality and a 40%increase in betweenness centrality for critical network nodes after multi-iteration optimization by using the proposed method.The shortest path length of network nodes decreases significantly,while clustering coefficient distribution becomes more balanced.The optimization process employs neighborhood information aggregation of nodes and dynamic weight adjustment as constraints,demonstrating the method's effectiveness in provincial transportation network optimization via enhancing network connectivity and robustness.关键词
综合立体交通/图神经网络/节点强化/网络优化/网络连通性Key words
Comprehensive Three-Dimensional Transportation/Graph Neural Network/Node Enhancement/Network Optimization/Network Connectivity分类
交通运输引用本文复制引用
王兆川,李舒霞,蒋军,罗斌文..基于图神经网络节点强化的四川省综合立体交通网络优化研究[J].铁道运输与经济,2025,47(5):96-107,12.基金项目
四川省青年科技创新研究团队项目(2023JBKY05) （2023JBKY05）
四川省科技厅项目(2023JDR0061) （2023JDR0061）
重庆市教育委员会人文社会科学研究项目(23SKGH141) （23SKGH141）
重庆市研究生联合培养基地课题(JDLHPYJD2022002) （JDLHPYJD2022002）
