摘 要：宏观基本图可从宏观层面对路网交通状态进行判别，但只能判别路网交通是否处于过饱和状态，无法进一步划分路网交通状态，而谱聚类算法能有效地对样本数据进行分类。鉴于此，文章提出基于宏观基本图和谱聚类的路网交通状态判别方法，首先利用浮动车估测法得到路网宏观基本图，然后依据谱聚类算法，对路网宏观基本图进行聚类分析，从而将路网交通状态划分为四个等级（低峰、平峰、高峰、过饱和）。为了验证算法的有效性，以广州市越秀区北京路周边路网为例，建立基于VISSIM 的微观交通仿真模型，设定浮动车比例为 30%，估测路网宏观基本图，然后在 MATLAB 中实现谱聚类算法，将路网交通划分为四种交通状态。研究的算法能够有效地对路网交通状态进行细分，为后续对路网交通进行精细化管理与控制奠定了基础。
关键词：交通工程；路网交通状态判别；宏观基本图；谱聚类；VISSIM 交通仿真
中图分类号：TP391.4 文献标识码：A 文章编号：2096-4706（2020）04-0113-03
Traffic State Discrimination Method and Simulation of Road Network Based on MFD and Spectral Clustering
WEI Wenyu，OUYANG Zhihao，LIU Qisen，LIANG Yilong，CAI Lichun
（Institute of Rail Traffic，Guangdong Communication Polytechnic，Guangzhou 510650，China）
Abstract：Macroscopic Fundamental Diagrams can distinguish the traffic state of road network from macroscopic level，but can only judge whether the road network traffic is too saturated，can not further divide the traffic state of the road network，and the spectral clustering algorithm can effectively classify the sample data. In view of this，this paper proposes a road network traffic state discrimination method based on macroscopic basic graph and spectral clustering，first using the floating vehicle estimation method to obtain the macroscopic basic map of the road network，and then according to the spectral clustering algorithm，the macroscopic basic map of the road network is cluster analysis，so as to divide the traffic state of the road network into four grades（low peak，flat peak，peak，supersaturation）. In order to verify the effectiveness of the algorithm，taking the road network around Beijing Road in Guangzhou Yuexiu District as an example，using VISSIM traffic simulation software，the simulation model of regional road network micro-traffic is established，and the proportion of floating vehicle is set as 30%，estimate the macroscopic basic map of the road network，and then realize the spectral clustering algorithm in MATLAB，which divides the road network into 4 kinds of traffic states. This algorithm can effectively subdivide the traffic state of the road network，which lays a foundation for the fine management and control of road network traffic in the next way.
Keywords：traffic engineering；road network traffic status discrimination；macroscopic fundamental diagrams；spectral clustering；VISSIM traffic simulation
基金项目：2020 年校级大学生科技创新项目（GDCP-ZX-2019-027-N2，GDCP-ZX-2019-025-N2）；2020 年广东大学生科技创新培育专项资金项目（pdjh202b0980）
参考文献：
[1] NAGLE A，GAYAH V. Accuracy of Networkwide Traffic States Estimated from Mobile Probe Data [J].Transportation Research Record Journal of the Transportation Research Board，2014（2421）：1-11.
[2] 林晓辉，徐建闽 . 基于自适应加权平均的路网 MFD 估测融合方法 [J]. 交通运输系统工程与信息，2018，18（6）：102-109.
[3] 刘仲民 . 基于图论的图像分割算法的研究 [D]. 兰州：兰州理工大学，2018.
[4] 商强，林赐云，杨兆升，等 . 基于谱聚类与 RS-KNN 的城市快速路交通状态判别 [J]. 华南理工大学学报（自然科学版），2017，45（6）：52-58.
作者简介：魏文钰（1999.01-），女，汉族，广东广州人，研究方向：交通控制与仿真。