基于支持向量机的城市道路交通状态判别方法研究

发布时间：2018-05-24 05:04

本文选题：交通状态判别 + 支持向量机　；参考：《青岛科技大学》2014年硕士论文

【摘要】：道路交通状态判别是智能交通管理系统进行路况信息发布和交通诱导的基础，目前我国各大城市的交通基础信息采集系统和交通诱导信息发布系统建设均已初具规模。然而在数据处理和实时道路交通状态判别中却还存在一些问题，亟待改进。交通状态判别是交通诱导中一个很重要的问题，提供实时准确的交通状态信息能为出行者做出合理的路径选择提供参考。因此，需要对路网中的交通状态信息做出准确判别预测。本文重点研究交通状态判别问题。对于交通状态判别问题，考虑到传统的交通状态划分方法是根据国家交通部门给定的交通状态指标来进行划分，这种方法对于不同道路的适应性不是很强，在实际应用中很难满足准确性要求。因此，本文采用模糊聚类的方法来划分交通状态，，对不同路段做出不同的状态划分。在此基础上，采用多类支持向量机方法对未来时刻的交通状况进行分类。多类支持向量机是传统两类支持向量机的改进，能够对具有多种划分指标的问题进行很好的划分，适合进行交通状态判别。本文提出了实时的交通状态判别系统模型。系统根据实时采集的交通参数信息准确的判断未来时刻的交通状态信息，并及时地将交通状态信息进行发布，很好地实现了交通诱导，提高了交通系统的服务质量。
[Abstract]:Road traffic condition identification is the basis of traffic information release and traffic guidance in intelligent traffic management system. At present, the construction of traffic basic information collection system and traffic guidance information publishing system in major cities in China has begun to take shape. However, there are still some problems in data processing and real-time traffic condition discrimination, which need to be improved. Traffic condition discrimination is an important problem in traffic guidance. Providing real-time and accurate traffic state information can provide a reference for travelers to make reasonable path selection. Therefore, it is necessary to make accurate discrimination and prediction of traffic state information in road network. This paper focuses on the problem of traffic state discrimination. For the problem of traffic condition discrimination, considering that the traditional traffic state classification method is based on the traffic state index given by the national transportation department, the adaptability of this method to different roads is not very strong. It is difficult to meet the requirement of accuracy in practical application. Therefore, this paper uses fuzzy clustering method to divide traffic state and make different state partition for different road sections. On this basis, multi-class support vector machine (SVM) method is used to classify traffic conditions in the future. Multi-class support vector machine (SVM) is an improvement of traditional two kinds of SVM. It can well divide the problems with multiple partitioning indexes and is suitable for traffic condition discrimination. In this paper, a real-time traffic condition discriminant system model is proposed. The system can accurately judge the traffic state information of the future time according to the traffic parameter information collected in real time, and publish the traffic state information in time, which realizes the traffic guidance and improves the service quality of the traffic system.
【学位授予单位】：青岛科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U491;TP181

【参考文献】