基于聚类的城市交通路网分区和交通状态判别

发布时间：2018-05-09 02:13

本文选题：聚类 + 路网分区　；参考：《北京交通大学》2017年硕士论文

【摘要】：近年来,随着城市化水平的不断提高,汽车保有量的不断增加,交通拥堵、出行安全、环境污染等一系列问题渐渐凸显。单纯增加城市交通基础设施的做法不仅耗费巨大的人力、物力,而且效果有限,因此如何对交通进行有效的管理和控制,是智能交通系统的重要研究内容。城市交通路网是一个规模巨大的复杂非线性时变系统,很难进行统一的管理和控制。对城市路网进行分区后,可以对每个子区域实施有针对性的控制方案,使路网系统变得灵活、可靠,而且需要实时处理的数据量明显减少,能够满足实时性的要求。本文选定望京区域作为研究对象,搭建VISSIM仿真平台,基于北京市浮动车数据设置仿真平台相关参数,复现研究区域的通行情况。之后利用聚类算法对望京区域进行分区,并利用宏观基本图(Macroscopic Fundamental Diagrams,MFD)从定性和定量的角度对分区的结果进行评价。本文的主要内容如下:首先,只考虑速度和密度信息,用K均值(K-means)聚类算法对路网进行分区,然后加入路段的空间位置信息再次进行分区,并将两种方法进行了对比。聚类结果表明本文提出的考虑路段空间位置的分区方法效果更好。并利用分区后子区域的MFD对K-means算法的分区结果进行评价,而且通过MFD的拟合函数,提出了一种衡量分区结果的标准。其次,使用改进的模糊C均值(FuzzyC-means,FCM)聚类算法确定聚类类数和初始聚类中心,对望京区域路网进行分区。利用分区后子区域的MFD评价改进的FCM算法的分区结果,评价结果表明,与K-means算法的分区结果相比,改进的FCM算法的分区结果更加理想。最后,用模糊综合评价方法对子区域内路段的交通状态进行判别,由此得出子区域内部拥堵路段比例随时间的变化情况,结合MFD的性质提出了一种子区域拥堵的评价标准。对望京区域路网进行动态分区,并分析了交通拥堵随时间的变化情况,确定了关键路段。
[Abstract]:In recent years, with the continuous improvement of the urbanization level, a series of problems such as traffic congestion, traffic congestion, travel safety and environmental pollution have become increasingly prominent. The practice of increasing urban traffic infrastructure only takes huge manpower, material resources and limited effect, so how to manage and control traffic effectively, It is an important research content of intelligent transportation system. The urban traffic network is a large and complex nonlinear time-varying system, it is difficult to carry out unified management and control. After the zoning of the urban road network, it can carry out a targeted control scheme for each subregion, making the network system flexible, reliable, and need to be processed in real time. In this paper, the Wangjing area is selected as the research object, and the VISSIM simulation platform is set up. Based on the parameters of the simulation platform of the floating car in Beijing, the current situation of the research area is recounted. Then the clustering algorithm is used to partition the Wangjing region, and the macro basic map (Mac Roscopic Fundamental Diagrams, MFD) evaluates the results of the partition from a qualitative and quantitative perspective. The main contents of this paper are as follows: first, we only consider the speed and density information, partition the road network with the K mean (K-means) clustering algorithm, and then partition the spatial location information of the section again, and carry out the two methods. The clustering results show that the partition method considering the location of the section is better. And the partition results of the K-means algorithm are evaluated by the MFD of the subregion subregion, and a criterion to measure the partition results is proposed through the fitting function of the MFD. Secondly, the improved fuzzy C mean (FuzzyC-means, FCM) clustering is used. The algorithm determines the number of clustering classes and the initial cluster center to partition the Wangjing regional road network. The results of the improved FCM algorithm are evaluated using the MFD of the subregion. The results show that the result of the improved FCM algorithm is more ideal than the K-means algorithm. Finally, the fuzzy comprehensive evaluation method is used to the subregion. The traffic state of the inner section is judged, thus the change of the proportion of the congested sections in the subregion is obtained with the change of the time. According to the nature of the MFD, the evaluation standard of a seed area congestion is put forward. The dynamic zoning of the Wangjing regional road network is carried out, and the change of traffic congestion with the time is analyzed, and the key sections are determined.

【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U12

【相似文献】