基于动态交通信息检测的干道交通拥堵预警方法研究
发布时间：2018-09-12 07:36
【摘要】：随着城市的不断发展，城市交通的矛盾主要表现为城市交通供给不能满足日益增长的交通需求，城市道路的交通拥堵问题越来越严重，逐渐成为制约城市和谐发展的一个全球性社会问题，缓解城市道路交通拥堵的工作显得越来越重要。城市干道是城市交通的动脉，对城市干道交通拥堵状况进行及时、准确的预测和识别，有针对性地对拥堵点采取交通控制和诱导等措施，可缓解干道交通瓶颈的拥堵程度，减少交通拥堵带来的负面效应。因此对城市干道交通拥堵预警建立科学有效的方法具有重要的实用价值。 本文利用交通流模型对城市干道交通拥堵在形成、持续、消散过程中流量、速度和密度之间的变化特性和时空特性进行了分析，提出交通拥堵预警是通过预测未来时刻干道某一截面的交通流状态参数，识别出该截面未来时刻的交通拥堵状况，预先采取有针对性的缓堵策略。在分析了交通流状态参数的基础上，对比分析了几种主要的交通信息数据采集技术，并对预处理交通信息数据的方法进行了探讨。 城市干道交通状态与相邻截面的交通状态密切相关，本文提出了基于多点状态参数的交通拥堵预警方法，建立预测的关键截面与相关的多个检测截面的交通状态相关模型，通过现状和历史交通流状态参数数据序列预测下一时段关键截面的交通流状态。采用ARIMA时间序列预测模型和遗传算法改进优化后的BP神经网络预测方法，建立了交通状态参数预测的线性组合模型，提出了基于最小误差平方和、等权和熵值法求解组合模型的权重值，并通过算例验证了基于最小误差平方和求权重的预测方法的预测效果最优。 最后，以平均速度、饱和度和平均延误作为城市干道交通拥堵的评价指标，将交通拥堵划分为畅通、轻微拥堵、拥堵和严重拥堵四个等级。提出对交通状态参数的预测结果采用改进的模糊综合评价方法，实现对交通拥堵程度的识别，并发出交通拥堵预警信息，，从而达到对城市干道交通拥堵的预警。 论文提出的交通拥堵预警方法能够及时有效的预警交通拥堵的发生，可用于智能交通系统中的交通状态预警和交通诱导。
[Abstract]:With the continuous development of the city, the contradiction of urban traffic mainly shows that the supply of urban traffic can not meet the increasing traffic demand, and the traffic congestion problem of urban roads is becoming more and more serious. It has gradually become a global social problem that restricts the harmonious development of cities, and it is becoming more and more important to alleviate urban road traffic congestion. Urban trunk roads are the arteries of urban traffic. Timely, accurate prediction and identification of traffic congestion on urban trunk roads, and targeted traffic control and guidance measures can alleviate the congestion degree of traffic bottlenecks in trunk roads. Reduce the negative effects of traffic congestion. Therefore, it is of great practical value to establish a scientific and effective method for traffic congestion warning on urban trunk roads. In this paper, traffic flow model is used to analyze the changing characteristics and space-time characteristics of traffic congestion between flow, velocity and density in the process of formation, persistence and dissipation of urban trunk road traffic congestion. It is proposed that traffic congestion warning is based on the prediction of traffic flow state parameters of a certain section of the main road in the future, the identification of traffic congestion at the future time of the section, and the adoption of a targeted strategy of slowing down the traffic congestion in advance. Based on the analysis of traffic flow state parameters, several main traffic information data acquisition techniques are compared and analyzed, and the methods of preprocessing traffic information data are discussed. The traffic state of urban trunk roads is closely related to the traffic state of adjacent sections. A traffic congestion warning method based on multi-point state parameters is proposed in this paper. The current and historical traffic flow state data series are used to predict the traffic flow state of critical sections in the next period. Using ARIMA time series prediction model and genetic algorithm to improve the optimized BP neural network prediction method, the linear combination model of traffic state parameter prediction is established, and the minimum error square sum is proposed. The method of equal weight and entropy is used to solve the weight value of the combined model, and an example is given to verify the optimal prediction effect of the prediction method based on the sum of least error square. Finally, with the average speed, saturation and delay as the evaluation index of traffic congestion, traffic congestion is divided into four grades: unblocked, slight, congested and severely congested. In this paper, an improved fuzzy comprehensive evaluation method is proposed to predict the traffic state parameters, which can recognize the traffic congestion degree and issue the traffic congestion warning information, so as to achieve the traffic congestion warning of the urban trunk roads. The traffic congestion warning method proposed in this paper can alert traffic congestion in time and effectively, and can be used in traffic state early warning and traffic guidance in intelligent transportation system (its).
【学位授予单位】：重庆交通大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U495;U491.265
本文编号：2238342
[Abstract]:With the continuous development of the city, the contradiction of urban traffic mainly shows that the supply of urban traffic can not meet the increasing traffic demand, and the traffic congestion problem of urban roads is becoming more and more serious. It has gradually become a global social problem that restricts the harmonious development of cities, and it is becoming more and more important to alleviate urban road traffic congestion. Urban trunk roads are the arteries of urban traffic. Timely, accurate prediction and identification of traffic congestion on urban trunk roads, and targeted traffic control and guidance measures can alleviate the congestion degree of traffic bottlenecks in trunk roads. Reduce the negative effects of traffic congestion. Therefore, it is of great practical value to establish a scientific and effective method for traffic congestion warning on urban trunk roads. In this paper, traffic flow model is used to analyze the changing characteristics and space-time characteristics of traffic congestion between flow, velocity and density in the process of formation, persistence and dissipation of urban trunk road traffic congestion. It is proposed that traffic congestion warning is based on the prediction of traffic flow state parameters of a certain section of the main road in the future, the identification of traffic congestion at the future time of the section, and the adoption of a targeted strategy of slowing down the traffic congestion in advance. Based on the analysis of traffic flow state parameters, several main traffic information data acquisition techniques are compared and analyzed, and the methods of preprocessing traffic information data are discussed. The traffic state of urban trunk roads is closely related to the traffic state of adjacent sections. A traffic congestion warning method based on multi-point state parameters is proposed in this paper. The current and historical traffic flow state data series are used to predict the traffic flow state of critical sections in the next period. Using ARIMA time series prediction model and genetic algorithm to improve the optimized BP neural network prediction method, the linear combination model of traffic state parameter prediction is established, and the minimum error square sum is proposed. The method of equal weight and entropy is used to solve the weight value of the combined model, and an example is given to verify the optimal prediction effect of the prediction method based on the sum of least error square. Finally, with the average speed, saturation and delay as the evaluation index of traffic congestion, traffic congestion is divided into four grades: unblocked, slight, congested and severely congested. In this paper, an improved fuzzy comprehensive evaluation method is proposed to predict the traffic state parameters, which can recognize the traffic congestion degree and issue the traffic congestion warning information, so as to achieve the traffic congestion warning of the urban trunk roads. The traffic congestion warning method proposed in this paper can alert traffic congestion in time and effectively, and can be used in traffic state early warning and traffic guidance in intelligent transportation system (its).
【学位授予单位】：重庆交通大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U495;U491.265
【参考文献】
相关期刊论文 前10条
1 陈华友;熵值法及其在确定组合预测权系数中的应用[J];安徽大学学报(自然科学版);2003年04期
2 张敬淦;北京交通拥堵症结分析[J];城市问题;2004年05期
3 王晓原;吴磊;张开旺;张敬磊;;非参数小波算法的交通流预测方法[J];系统工程;2005年10期
4 杨兆升,杨庆芳,冯金巧;基于模糊综合推理的道路交通事件识别算法[J];公路交通科技;2003年04期
5 姜桂艳;常安德;牛世峰;;基于车牌识别数据的交通拥堵识别方法[J];哈尔滨工业大学学报;2011年04期
6 王宏杰,林良明,徐大淦,颜国正;基于改进BP网交通流动态时序预测算法的研究[J];交通与计算机;2001年03期
7 楚杨杰;陈春红;刘昭;王雄;宋兵;;道路交通拥堵自动检测的改进加州算法及仿真[J];数学杂志;2012年04期
8 罗小强;陈宽民;张同芬;;城市快速路交通拥挤事件检测与程度辨识[J];长安大学学报(自然科学版);2010年03期
9 吴志周;范宇杰;马万经;;基于灰色神经网络的点速度预测模型[J];西南交通大学学报;2012年02期
10 贺国光,马寿峰,李宇;基于小波分解与重构的交通流短时预测法[J];系统工程理论与实践;2002年09期
本文编号：2238342
本文链接：https://www.wllwen.com/kejilunwen/jiaotonggongchenglunwen/2238342.html
