高速公路交通流预测及事故预警方法研究
发布时间：2018-07-15 11:52
【摘要】：提高高速公路事故管理水平对于保障国民经济健康发展、人民出行安全舒适、社会生活平安和谐具有重要意义。本文首先从“危机管理”的角度对交通事故管理所涉及的各个阶段进行了分析，提出了高速公路交通事故管理系统的框架结构并对框架各部分功能进行了概述，总结了与各个基本功能部分有关的研究现状，并选择当前研究比较薄弱的“事故事前处理”部分作为本文研究的重点，按交通流预测—交通事故预警—事故提前应对的逻辑层次展开研究，主要研究内容和结果包括： 一：交通流短时预测方法和预测系统研究，首先提出了一种基于内聚度指标和遗传算法的状态向量选择算法，在四种不同的情景下对算法的效果进行了验证，证明本文提出的算法能够明显提高非参数回归预测和神经网络预测的精度。随后提出了一种数据驱动的交通流短时预测系统的架构，可以用于大路网、多点位实时预测。实际数据实验和仿真实验都表明本文提出的预测系统能够应对大路网、高数据流量条件下的预测，同时能取得较好的预测效果。 二：高速公路事故时空分布规律分析及事故预警算法研究。以津蓟高速的事故数据为基础，对高速公路交通事故时空分布规律进行了分析，发现高速公路上的交通事故与一天中的时段、日期、地点、车型比例和交通流量都有明显的关系，证明了高速公路事故的时空规律性。提出了一种基于模糊推理的高速公路交通事故的预警算法，通过对实时采集的交通流参数进行分析，识别交通流的异常状态，基于天津市实测数据的实验表明，该方法能够比较准确的预测交通事故风险，适合作为管理者决策参考。 三：高速公路事故预先应对方法。提出了一种高速公路巡逻车辆的优化调度算法。以车辆路径规划问题为基础，将预测的路段事故风险加入目标函数中，随后设计了一种两阶段的启发式算法进行求解，并在不同高速公路网络状态条件下进行了实验。证明该算法能够提高巡逻车辆对事故路段的覆盖率，提高事故反应时间。 综上所述，本文针对高速公路交通事故管理中的“事前处理”阶段展开研究，提出了涵盖交通流预测、事故预警、预警之后的应对处理等阶段的若干方法。基于实际数据和仿真软件的实验表明，，本文提出的方法对特定问题能取得较好的效果，具有一定的可行性。
[Abstract]:It is of great significance to improve the management level of expressway accidents to ensure the healthy development of national economy, the safety and comfort of people's travel, and the peace and harmony of social life. At first, this paper analyzes the stages of traffic accident management from the angle of "crisis management", puts forward the frame structure of expressway traffic accident management system, and summarizes the functions of each part of the frame. This paper summarizes the current research situation related to each basic functional part, and selects the "accident prior treatment" section, which is relatively weak in current research, as the focal point of this paper. According to the logical level of traffic flow forecasting, traffic accident warning and accident early response, the main research contents and results are as follows: first, traffic flow short-term forecasting method and forecasting system research, Firstly, a state vector selection algorithm based on cohesion index and genetic algorithm is proposed, and the effect of the algorithm is verified in four different scenarios. It is proved that the proposed algorithm can significantly improve the accuracy of nonparametric regression prediction and neural network prediction. Then, a data driven short time traffic flow prediction system is proposed, which can be used for real time prediction of large road networks and multi points. The actual data experiments and simulation experiments show that the proposed forecasting system can deal with the prediction under the condition of large road network and high data flow, and can achieve good prediction results. Second, the analysis of temporal and spatial distribution of expressway accidents and the research of accident warning algorithm. Based on the accident data of artichoke expressway, the temporal and spatial distribution of expressway traffic accidents is analyzed. It is found that the traffic accidents on the freeway are related to the period, date and place of the day. The proportion of vehicle type and traffic flow are obviously related, which proves the temporal and spatial regularity of highway accidents. This paper presents an early warning algorithm for freeway traffic accidents based on fuzzy reasoning. By analyzing the traffic flow parameters collected in real time, the abnormal state of traffic flow is identified. The experimental results based on the measured data in Tianjin show that, This method can accurately predict the risk of traffic accidents and is suitable for decision-making of managers. Three: highway accident in advance. This paper presents an optimal scheduling algorithm for highway patrol vehicles. Based on the vehicle path planning problem, the predicted road accident risk is added to the objective function, and then a two-stage heuristic algorithm is designed to solve the problem, and experiments are carried out under different expressway network conditions. It is proved that the algorithm can improve the coverage of patrol vehicles and improve the response time. To sum up, this paper studies the stage of "prior treatment" in expressway traffic accident management, and puts forward some methods including traffic flow prediction, early warning of accidents, and response to early warning. Experiments based on actual data and simulation software show that the method proposed in this paper can achieve good results for specific problems and has certain feasibility.
【学位授予单位】：天津大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：U491
本文编号：2123979
[Abstract]:It is of great significance to improve the management level of expressway accidents to ensure the healthy development of national economy, the safety and comfort of people's travel, and the peace and harmony of social life. At first, this paper analyzes the stages of traffic accident management from the angle of "crisis management", puts forward the frame structure of expressway traffic accident management system, and summarizes the functions of each part of the frame. This paper summarizes the current research situation related to each basic functional part, and selects the "accident prior treatment" section, which is relatively weak in current research, as the focal point of this paper. According to the logical level of traffic flow forecasting, traffic accident warning and accident early response, the main research contents and results are as follows: first, traffic flow short-term forecasting method and forecasting system research, Firstly, a state vector selection algorithm based on cohesion index and genetic algorithm is proposed, and the effect of the algorithm is verified in four different scenarios. It is proved that the proposed algorithm can significantly improve the accuracy of nonparametric regression prediction and neural network prediction. Then, a data driven short time traffic flow prediction system is proposed, which can be used for real time prediction of large road networks and multi points. The actual data experiments and simulation experiments show that the proposed forecasting system can deal with the prediction under the condition of large road network and high data flow, and can achieve good prediction results. Second, the analysis of temporal and spatial distribution of expressway accidents and the research of accident warning algorithm. Based on the accident data of artichoke expressway, the temporal and spatial distribution of expressway traffic accidents is analyzed. It is found that the traffic accidents on the freeway are related to the period, date and place of the day. The proportion of vehicle type and traffic flow are obviously related, which proves the temporal and spatial regularity of highway accidents. This paper presents an early warning algorithm for freeway traffic accidents based on fuzzy reasoning. By analyzing the traffic flow parameters collected in real time, the abnormal state of traffic flow is identified. The experimental results based on the measured data in Tianjin show that, This method can accurately predict the risk of traffic accidents and is suitable for decision-making of managers. Three: highway accident in advance. This paper presents an optimal scheduling algorithm for highway patrol vehicles. Based on the vehicle path planning problem, the predicted road accident risk is added to the objective function, and then a two-stage heuristic algorithm is designed to solve the problem, and experiments are carried out under different expressway network conditions. It is proved that the algorithm can improve the coverage of patrol vehicles and improve the response time. To sum up, this paper studies the stage of "prior treatment" in expressway traffic accident management, and puts forward some methods including traffic flow prediction, early warning of accidents, and response to early warning. Experiments based on actual data and simulation software show that the method proposed in this paper can achieve good results for specific problems and has certain feasibility.
【学位授予单位】：天津大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：U491
【参考文献】
相关期刊论文 前7条
1 刘荣;;我国高速公路管理体制存在的问题及对策研究[J];产业与科技论坛;2010年01期
2 李相勇,张南,蒋葛夫;道路交通事故灰色马尔可夫预测模型[J];公路交通科技;2003年04期
3 陈铁明;龚荣盛;SAMUEL H Huang;;一种新的快速模糊规则提取方法[J];控制与决策;2008年09期
4 孙才志,王敬东,潘俊;模糊聚类分析最佳聚类数的确定方法研究[J];模糊系统与数学;2001年01期
5 郑黎黎;丁同强;范海燕;杨兆升;孙健;;高速公路交通事件影响范围的模糊预测[J];数学的实践与认识;2009年01期
6 姜紫峰，韩锡令;高速公路交通流中等密度区的可变速度和入口匝道联合控制[J];西安公路交通大学学报;1994年04期
7 马寿峰,贺国光,刘豹;智能交通系统中短时交通流预测系统的研究[J];预测;2004年02期
本文编号：2123979
本文链接：https://www.wllwen.com/kejilunwen/jiaotonggongchenglunwen/2123979.html
