基于拥堵转移矩阵的城市拥堵路段预测方法研究

发布时间：2018-03-05 21:41

本文选题：突发拥堵路段　切入点：交通流转移概率　出处：《东北大学》2014年硕士论文　论文类型：学位论文

【摘要】：近年来,人口出生率和机动车占有率不断增加,城市机动车数量迅猛发展,道路资源建设相对滞后,导致城市交通拥堵日益严重,城市拥堵路段预测成为智能交通系统的一个基本功能。准确的预测城市拥堵路段,不仅可以使出行者更好的选择出行线路,而且可以减少尾气排放,达到节约时间、经济、能源,保护环境的目的。虽然对于拥堵路段预测的研究已经存在了相当长的一段时间,但是在速度、准确性、可扩展性等方面仍不完善,尤其是突发事件的出现使拥堵路段预测的准确性大大降低。为了解决上述问题,本文提出一种新的基于拥堵转移矩阵的城市拥堵路段预测模型,该模型与其他大部分交通流预测模型明显的不同是它结合了相邻路段的交通流信息进行拥堵路段的预测,得到较好的预测效果。本文的主要贡献包括：首先,针对拥堵路段间具有时空因果关系的特点,提出对城市拥堵路段预测建立数据模型的具体算法。将GPS轨迹集按照周期和时间框架分类,挖掘不同时间框架内的拥堵路段和交通流转移概率。其次,提出拥堵转移矩阵的概念。拥堵转移矩阵记录了具有时空因果关系的拥堵路段对之间的拥堵转移概率。为每一个时间框架建立一个拥堵转移矩阵,从而为拥堵路段的预测做准备。再次,提出对突发拥堵路段的处理方法。该方法利用突发拥堵路段和与其相邻路段之间的交通流转移概率预测下一时间框架内它们中可能发生拥堵的路段,提高预测准确率。最后,在全概率理论的基础上,提出利用拥堵向量和拥堵转移矩阵相乘的方法进行一步或多步拥堵路段的预测。通过实验评估,验证了本文提出的基于拥堵转移矩阵预测城市拥堵路段算法的有效性和高效性。
[Abstract]:In recent years, the birth rate and the share of motor vehicles have been increasing, the number of motor vehicles in urban areas has developed rapidly, and the construction of road resources has lagged behind, which has led to the increasingly serious traffic congestion in cities. Prediction of urban congested sections has become a basic function of its. Accurate prediction of urban congested sections can not only make travelers better choose travel routes, but also reduce exhaust emissions, thus saving time and economy. Energy, the purpose of protecting the environment. Although the research on the prediction of congested roads has existed for quite a long time, it is still imperfect in terms of speed, accuracy, extensibility, etc. In order to solve the above problems, a new urban congestion section prediction model based on congestion transfer matrix is proposed in this paper. The difference between this model and most other traffic flow forecasting models is that it combines the traffic flow information of adjacent road sections to predict congested sections and gets better prediction results. The main contributions of this paper are as follows: first of all, In view of the spatial-temporal causal relationship between congested sections, a data model for urban congested road sections is proposed. The GPS trajectory sets are classified according to the period and time frame. Excavate the probability of traffic flow transfer in different time frame. Secondly, The concept of congestion transfer matrix is proposed. Congestion transfer matrix records the probability of congestion transfer between congested road pairs with temporal and spatial causality, and establishes a congestion transfer matrix for each time frame. To prepare for the prediction of congested roads. Again, This method uses the probability of traffic flow transfer between sudden congested sections and its adjacent sections to predict the possible congested sections in the next time frame, and improves the accuracy of prediction. Based on the theory of full probability, the method of multiplying congestion vector and congestion transfer matrix is proposed to predict one-step or multi-step congested road sections. The effectiveness and efficiency of the proposed algorithm based on congestion transfer matrix are verified.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U491

【相似文献】