基于数据挖掘的交通流机理分析

发布时间：2018-03-21 13:24

本文选题：交通流　切入点：数据挖掘　出处：《华南理工大学》2014年硕士论文　论文类型：学位论文

【摘要】：城市快速路通过改善道路条件和交通环境,以大容量、高效率的交通功能为人们出行提供了快速、高效的运行环境,具有良好的社会效益和经济效益。目前，我国关于城市快速路的交通流基础理论研究正处于发展阶段，国内学者多利用数学模型来描述交通状态及其变化，并已取得一定成果。但在基于实证法的交通流研究中，国内研究还较少涉及。为此，论文结合数据挖掘和实证方法探究了城市快速路交通流运作机理，建立了基于交通参数统计特性、PCA和BP神经网络的交通相辨识与预测模型，为城市交通系统管理提供理论支持和应用参考。本文首先通过交通参数的时序图、基本图和时空数据重构法，定性分析交通流相态及其演变的数据特征和拥挤蔓延特性；随后论文利用统计原理，以Occ为自变量分析了Q、V、ht及其时间差分量的标准差变化特性，建立了Q-Occ判别相图，拟合出了F-S相变和S-J相变的S型概率曲线，实现了定性分析到定量分析的转化；最后，论文利用PCA和KNN方法在低维空间分析了交通相的数据特征，确定了Q、V、Occ、ht的特征指标，建立了基于PCA和神经网络的交通相辨识与预测模型，作为统计模型的补充。论文的创新点主要有：（1）根据城市快速路交通流实测数据，详细阐述了城市快速路交通流运行机理，利用统计原理建立了Q-Occ判别相图，拟合出了F-S相变和S-J相变的概率曲线；（2）通过PCA和KNN的方法确定了适用于特征提取的交通参数集合,设计了基于PCA特征提取的交通相辨识模型和BP神经网络的相演变预测模型。根据实测数据的检验，该模型能较好地判别出当前交通状态，并且能通过“三相匹配曲线”展现三相交通流的博弈过程。三相匹配曲线一定程度上量化了交通相的动态演变过程，并结合BP神经网络应用在交通相演变预测模型。论文中涉及到的交通流机理分析、交通相辨识模型和相演变预测模型，可为智能交通系统的控制策略提供一定的理论依据和应用参考。
[Abstract]:By improving the road conditions and traffic environment, urban expressway provides a fast and efficient environment for people to travel with large capacity and high efficiency. It has good social and economic benefits. The research on the basic theory of urban expressway traffic flow in China is in the developing stage. Domestic scholars often use mathematical models to describe the traffic state and its changes, and have achieved certain results. However, in the research of traffic flow based on empirical method, Therefore, the paper studies the operation mechanism of urban expressway traffic flow based on data mining and empirical methods, and establishes a traffic phase identification and prediction model based on PCA and BP neural network, which is based on the statistical characteristics of traffic parameters. It provides theoretical support and application reference for urban traffic system management. In this paper, firstly, we qualitatively analyze the data characteristics of traffic flow phase and its evolution and congestion spread by using the sequential diagram of traffic parameters, the basic map and the method of reconstruction of spatiotemporal data, and then use the statistical principle to analyze the phase behavior of traffic flow and the characteristics of traffic congestion spread. Using Occ as an independent variable, the variation characteristics of the standard deviation of Q / V _ (t) and its time difference component are analyzed, and the Q-Occ discriminant phase diagram is established, and the S-type probability curves of F-S phase transition and S-J phase transition are fitted out, and the transformation from qualitative analysis to quantitative analysis is realized. In this paper, the PCA and KNN methods are used to analyze the data features of traffic phase in low dimensional space, and the characteristic indexes of QTV-Occht are determined. A traffic phase identification and prediction model based on PCA and neural network is established, which is used as a supplement to the statistical model. The innovations of this paper are as follows:. 1) based on the measured data of urban expressway traffic flow, the operation mechanism of urban expressway traffic flow is expounded in detail, the Q-Occ discriminant phase diagram is established by using the statistical principle, and the probability curves of F-S phase transition and S-J phase transition are fitted out. (2) the set of traffic parameters suitable for feature extraction is determined by means of PCA and KNN, and the traffic phase identification model based on PCA feature extraction and the phase evolution prediction model of BP neural network are designed. The model can distinguish the current traffic state and show the game process of the three-phase traffic flow through "three-phase matching curve". To some extent, the three-phase matching curve quantifies the dynamic evolution process of the traffic phase. Combined with BP neural network, it is applied to the prediction model of traffic phase evolution. The analysis of traffic flow mechanism, traffic phase identification model and phase evolution prediction model in this paper can provide some theoretical basis and application reference for intelligent transportation system control strategy.
【学位授予单位】：华南理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U491.112

【参考文献】