基于车牌跟踪的交通运行状态评价及预测

发布时间：2018-04-10 03:07

本文选题：交通运行状态　切入点：车牌跟踪数据　出处：《华南理工大学》2014年硕士论文

【摘要】：城市道路交通运行状态的准确实时评价及预测是准确把握城市道路交通系统行为、科学制定交通管理决策的基础，如何定量描述交通运行状态以及如何面向交通参与者对交通状态进行不同层次的评价和预测,一直是城市交通研究领域的重点和难点,对这些问题的研究具有重要的理论和现实意义。本论文以城市交通管理的实际需求为背景,在假定车牌数据可以获取以及对城市道路交通流特性进行分析的基础上，构建起一套适用于城市道路交通系统的完整可操作的交通运行状态评价及预测体系，主要研究成果具体体现为以下几个方面:给出基于出行者心理考虑的交通运行状态的定义，并据此提出了路段单元、OD路径和路网三个层面上的交通行程指数作为城市交通运行状态的评价指标，验证了指标的有效性，并对评价单元、更新步长和状态划分标准进行了研究，进而设计了基于模糊评判的多参数融合评价模型；对车牌数据的处理分析过程进行了研究，针对城市道路单车行程行程时间波动大的特点，设计了基于时间窗格划分和数据序列分位点值截取的行程时间异常数据过滤方法，并对主要数据表格进行了设计；提出了一种基于城市交通特性和时空二维影响因素的短时交通参数组合预测模型，并用实测的交通流数据进行了验证，结果表明该模型具有较好的预测精度和对不同交通运行状态的适应性；改进了基于预测偏差的交通事件预警方法，加入目标路段单元与上下游主要流入的路段单元的预测偏差值符号判断的步骤，旨在进一步提高交通事件检测的准确性；对车牌数据的处理分析流程和实时交通运行状态评价的可视化进行了仿真及演示。通过对基于android的手持式交通调查仪获取的实测数据和基于VISSIM仿真的数据进行的处理分析验证了论文构建的城市道路交通运行状态评价及预测体系的有效性，，为后续的相关研究提供了思路方法。
[Abstract]:The accurate real-time evaluation and prediction of urban road traffic running state is the basis of accurately grasping the behavior of urban road traffic system and making traffic management decision scientifically.How to quantitatively describe the traffic running state and how to evaluate and predict the traffic state at different levels for traffic participants has always been the focus and difficulty in the field of urban traffic research.The study of these problems has important theoretical and practical significance.This paper takes the actual demand of urban traffic management as the background, on the basis of assuming that the license plate data can be obtained and analyzing the characteristics of urban road traffic flow.A set of integrated and operable traffic operation evaluation and prediction system is constructed for urban road traffic system. The main research results are as follows: the definition of traffic running state based on travelers' psychological considerations is given.On the basis of this, the OD path of road section unit and the traffic travel index at three levels of road network are put forward as the evaluation index of urban traffic running state, and the validity of the index is verified, and the evaluation unit is given.The updated step size and state partition criteria are studied, and then a multi-parameter fusion evaluation model based on fuzzy evaluation is designed, and the processing and analysis process of license plate data is studied.Aiming at the large fluctuation of cycle travel time on urban roads, a method of filtering abnormal travel time data based on time pane partition and data sequence locus interception is designed, and the main data tables are designed.A combined forecasting model of short-term traffic parameters based on the characteristics of urban traffic and two-dimensional influence factors of time and space is proposed and verified by the measured traffic flow data.The results show that the model has good prediction accuracy and adaptability to different traffic running states, and improves the traffic event warning method based on forecasting deviation.In order to improve the accuracy of traffic incident detection, the step of judging the prediction deviation value symbol between the target road section unit and the upstream and downstream main section unit is added to further improve the accuracy of traffic incident detection.The simulation and demonstration of the processing and analyzing flow of license plate data and the visualization of real-time traffic running state evaluation are carried out.The validity of the evaluation and prediction system of urban road traffic running state is verified by processing and analyzing the measured data obtained by the hand-held traffic survey instrument based on android and the data based on VISSIM simulation.It provides a way of thinking for the following related research.
【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

【参考文献】