高速公路多源异构交通数据融合与预测方法研究

发布时间：2018-03-22 22:10

本文选题：交通信息融合　切入点：固定采集　出处：《吉林大学》2015年硕士论文　论文类型：学位论文

【摘要】：随着我国经济快速的发展，人口的增加以及收入的增长，汽车的拥有量激增，使得汽车的出行比例越来越高，城市交通面临着巨大的挑战。交通拥堵、交通污染、交通安全问题频繁发生，严重影响了人们的正常生活，，也对我国国民经济的持续健康发展提出了挑战。我国交通的资料比较少，所以很多城市因为资料的匿乏造成了交通管理、规划和控制等各方面的困难。交通管理需要大量的交通数据，但是单一的检测器获得的交通数据显然不能满足它对交通数据的需求。随着信息融合技术的诞生并且飞速发展，这一问题能够被有效的解决。为了实现对交通状态进行动态估计，必须首先对路网绝大多数（甚至是全部）路段的交通参数进行估计。然而，目前在我国（甚至是发达国家）的高速公路中并不是所有的路段都安装有检测装置。按照以上所述问题，本文的主要研究内容是高速公路多源异构交通数据融合与预测方法，首先对多源交通信息质量评价与控制技术进行研究，提出了高速公路中固定采集与移动采集的融合方法、基于历史数据与实时数据的交通流预测方法。固定采集与移动采集的融合方法。本文介绍了最小方差加权平均法的基本原理以及权值的确定方法。因为目前在高速公路中采集交通信息主要以视频装置和浮动车为主，所以本文以区间平均速度为例介绍了基于自适应加权平均法的固定采集与移动采集的快速融合方法。最后设计了融合方法的功能模块并对提出的方法进行了实例验证，通过实验证明了本文提出方法的有效性。基于历史数据与实时数据的交通流预测方法。由于技术和管理的限制，高速公路中检测器经常毁坏或者一些路段根本没有安装检测器，导致一些路段没法检测交通数据，有时也因为天气的原因导致有检测器路段检测交通参数不准确。本文先利用聚类分析法分析无检测器路段以及相邻的路段的历史数据，利用历史数据和实时数据对交通流进行预测。然后基于交通流理论利用无检测器路段的浮动车实时数据和上游路段的历史数据对交通流进行预测。最后对功能模块进行了设计并且进行了实验验证。
[Abstract]:With the rapid development of economy in our country, the increase of population and income growth, a surge in the amount of the car, making the car travel more and more, city traffic is facing a huge challenge. Traffic congestion, traffic pollution, frequent traffic safety problems, seriously affecting people's normal life, but also a challenge of China's national economy sustained and healthy development of China's transportation. The data is relatively small, so the lack of data because many of the city caused traffic management, all aspects of planning and control difficulties. Traffic management requires a large amount of data traffic, but the traffic data obtained from single detector cannot satisfy its traffic data the demand of information fusion technology. With the birth and rapid development, this problem can be solved effectively. In order to realize the dynamic estimation of the traffic state, must first vast road network Most (or all) sections of the traffic parameters estimation. However, at present in our country (even developed countries) of the highway and not all sections are installed in a detection device. According to the above mentioned problems, the main content of this paper is the highway of multi-source and heterogeneous traffic data fusion and prediction method research the first technique and control of multi-source traffic information quality evaluation, proposed fusion method of fixed acquisition and mobile acquisition of expressway, prediction method of historical data and real-time data based on traffic flow.
Fusion method of acquisition and fixed mobile acquisition. This paper introduces the method of determining the minimum variance weighted average method and the principle of weight. Because the current highway traffic information collected in the video device and the floating car, so this paper takes segmentspeed as an example to introduce the rapid integration of fixed and mobile data acquisition method of adaptive weighted based on the averaging method. The function module design of the fusion method and the proposed method is verified by experiment, proved the effectiveness of the proposed method.
The prediction method of historical data and real-time data of traffic flow based on technology and management. Due to restrictions in highway detector often destroyed or some sections did not install detector, cause some sections can not detect traffic data, sometimes because of the weather caused a traffic parameter detection detector section is not accurate. This paper first use clustering analysis method the analysis of historical data and the non detector section adjacent to the road, to predict traffic flow by using historical data and real-time data. Then the traffic flow history data by using the theory of non detector section of the floating car data and upstream section on traffic flow prediction based on the end of the function module has been designed and tested.

【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U491

【参考文献】