城市道路短时交通流动态预测方法研究与应用
发布时间：2019-04-24 01:16
【摘要】：实时、准确、动态的短时交通流预测一直是智能交通发展极力追求的目标。然而,短时交通流信息量大,受到不确定噪声信号干扰强,再加上城市路网复杂的拓扑结构,导致如何实现城市道路短时交通流动态预测这一问题一直制约着智能交通长足的发展。为了解决上述问题许多预测方法相继被提出,但其中一些方法都因未考虑短时交通流不确定干扰因素或者城市路网复杂性的影响,致使预测结果实时性与准确性都不尽理想,也没有达到实际意义上的动态预测。本文在Mallat算法下对短时交通流信号进行小波分解与重构,在滤掉短时交通流信息强干扰噪声信号后进行时、频域特性分析,大大提高了短时交通流信息预处理速度及精度,再将支持向量回归预测模型结合在城市路网多断面的预测思想中,使得城市道路短时交通流动态预测更加准确、实时、有效。论文旨在研究短时交通流有效信息快速、有效提取后进行城市路网中短时交通流精确的动态预测。论文主要研究工作如下:1.分析了交通流、速度、密度三个短时交通流基本参数的关系及数学模型;归纳了短时交通流基本特性;从空间、时间的层面出发,探讨了短时交通流的相关性;从动力学特性角度总结了短时交通流可预测性分析方法及其影响因素。2.针对短时交通流信号的强噪声影响,研究了小波分解与重构对短时交通流信号进行主体信息和细节信息快速、有效提取的关键问题。通过Mallat塔式多分辨率算法思想的引入,实现了短时交通流信号的快速分解与重构,最后提出了针对短时交通流信号如何有效提取的详细解决策略。3.以实现短时交通流动态预测为出发点,给出原始交通流数据预处理的方法,并通过SVR自身特性的研究,提出了模型自适应参数、G-P算法嵌入维数与核函数的动态优化与选择方法,使得短时交通流达到高效、准确的动态预测。4.通过对城市复杂网络拓扑结构的研究,给出了相关断面的矩阵表示方法及相互影响权重的F-AHP模糊标定方法;最后结合前面的研究结果提出了一个城市道路短时交通流动态预测模型,并通过实验验证了其合理性。5.利用短时交通流信号的Mallat小波分解与重构及SVR预测的研究结论,在ThinkPHP框架下设计并完成了针对西安市的短时交通流动态预测系统的开发。
[Abstract]:Real-time, accurate and dynamic short-term traffic flow prediction is always the goal of intelligent transportation development. However, short-term traffic flow has a large amount of information and is strongly disturbed by uncertain noise signals, coupled with the complex topological structure of the urban network. The problem of how to realize the dynamic prediction of urban road short-term traffic flow has been restricting the rapid development of intelligent traffic. In order to solve the above problems, many forecasting methods have been put forward one after another, but some of these methods have not considered the influence of uncertain disturbance of short-term traffic flow or the complexity of urban road network, so that the real-time and accuracy of prediction results are not ideal. Also did not achieve the actual meaning of the dynamic prediction. In this paper, the short-time traffic flow signal is decomposed and reconstructed by wavelet transform based on Mallat algorithm. When the short-time traffic flow information is filtered out, the frequency domain characteristic is analyzed, which greatly improves the speed and precision of short-time traffic flow information preprocessing. Then the support vector regression prediction model is combined with the multi-section prediction idea of urban road network, which makes the dynamic prediction of urban road short-term traffic flow more accurate, real-time and effective. The purpose of this paper is to study the short-term traffic flow effective information quickly, and then extract the short-term traffic flow accurately and dynamically forecast the short-term traffic flow in the urban road network. The main research work of this paper is as follows: 1. This paper analyzes the relationship and mathematical model of three basic parameters of short-term traffic flow, such as traffic flow, velocity and density, sums up the basic characteristics of short-term traffic flow, discusses the correlation of short-term traffic flow from the aspect of space and time, discusses the relationship between short-term traffic flow and short-term traffic flow. The predictive analysis method of short-term traffic flow and its influencing factors are summarized from the point of view of dynamic characteristics. Aiming at the influence of strong noise on short-time traffic flow signal, the key problem of fast and efficient extraction of main body information and detail information of short-time traffic flow signal based on wavelet decomposition and reconstruction is studied. Through the introduction of Mallat tower multi-resolution algorithm, the fast decomposition and reconstruction of short-time traffic flow signal is realized. Finally, a detailed solution strategy for how to extract the short-time traffic flow signal effectively is put forward. 3. In order to realize the dynamic prediction of short-term traffic flow, the method of pre-processing the original traffic flow data is given, and the adaptive parameters of the model are put forward through the study of the characteristics of SVR itself. The dynamic optimization and selection method of embedding dimension and kernel function makes the short-term traffic flow achieve efficient and accurate dynamic prediction. 4. Based on the study of the topological structure of urban complex network, the matrix representation method of correlation cross-section and the fuzzy calibration method of F-AHP for mutual influence weights are given. Finally, a dynamic prediction model of urban road short-term traffic flow is proposed based on the previous research results, and its rationality is verified by experiments. Based on the research results of Mallat wavelet decomposition and reconstruction of short-term traffic flow signals and SVR prediction, a short-term traffic flow dynamic prediction system for Xi'an is designed and completed under the framework of ThinkPHP.
【学位授予单位】：长安大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U491.14
本文编号：2463980
[Abstract]:Real-time, accurate and dynamic short-term traffic flow prediction is always the goal of intelligent transportation development. However, short-term traffic flow has a large amount of information and is strongly disturbed by uncertain noise signals, coupled with the complex topological structure of the urban network. The problem of how to realize the dynamic prediction of urban road short-term traffic flow has been restricting the rapid development of intelligent traffic. In order to solve the above problems, many forecasting methods have been put forward one after another, but some of these methods have not considered the influence of uncertain disturbance of short-term traffic flow or the complexity of urban road network, so that the real-time and accuracy of prediction results are not ideal. Also did not achieve the actual meaning of the dynamic prediction. In this paper, the short-time traffic flow signal is decomposed and reconstructed by wavelet transform based on Mallat algorithm. When the short-time traffic flow information is filtered out, the frequency domain characteristic is analyzed, which greatly improves the speed and precision of short-time traffic flow information preprocessing. Then the support vector regression prediction model is combined with the multi-section prediction idea of urban road network, which makes the dynamic prediction of urban road short-term traffic flow more accurate, real-time and effective. The purpose of this paper is to study the short-term traffic flow effective information quickly, and then extract the short-term traffic flow accurately and dynamically forecast the short-term traffic flow in the urban road network. The main research work of this paper is as follows: 1. This paper analyzes the relationship and mathematical model of three basic parameters of short-term traffic flow, such as traffic flow, velocity and density, sums up the basic characteristics of short-term traffic flow, discusses the correlation of short-term traffic flow from the aspect of space and time, discusses the relationship between short-term traffic flow and short-term traffic flow. The predictive analysis method of short-term traffic flow and its influencing factors are summarized from the point of view of dynamic characteristics. Aiming at the influence of strong noise on short-time traffic flow signal, the key problem of fast and efficient extraction of main body information and detail information of short-time traffic flow signal based on wavelet decomposition and reconstruction is studied. Through the introduction of Mallat tower multi-resolution algorithm, the fast decomposition and reconstruction of short-time traffic flow signal is realized. Finally, a detailed solution strategy for how to extract the short-time traffic flow signal effectively is put forward. 3. In order to realize the dynamic prediction of short-term traffic flow, the method of pre-processing the original traffic flow data is given, and the adaptive parameters of the model are put forward through the study of the characteristics of SVR itself. The dynamic optimization and selection method of embedding dimension and kernel function makes the short-term traffic flow achieve efficient and accurate dynamic prediction. 4. Based on the study of the topological structure of urban complex network, the matrix representation method of correlation cross-section and the fuzzy calibration method of F-AHP for mutual influence weights are given. Finally, a dynamic prediction model of urban road short-term traffic flow is proposed based on the previous research results, and its rationality is verified by experiments. Based on the research results of Mallat wavelet decomposition and reconstruction of short-term traffic flow signals and SVR prediction, a short-term traffic flow dynamic prediction system for Xi'an is designed and completed under the framework of ThinkPHP.
【学位授予单位】：长安大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U491.14
【参考文献】
相关期刊论文 前2条
1 陈岳明;萧德云;;基于动态交通分配的路网应急疏散模型[J];清华大学学报(自然科学版);2009年08期
2 孙湘海;刘潭秋;;基于神经网络和SARIMA组合模型的短期交通流预测[J];交通运输系统工程与信息;2008年05期
相关博士学位论文 前2条
1 姚智胜;基于实时数据的道路网短时交通流预测理论与方法研究[D];北京交通大学;2007年
2 崔立成;基于多断面信息的城市道路网交通流预测方法研究[D];大连海事大学;2012年
相关硕士学位论文 前1条
1 曾嵘;基于时空关联状态的多路段断面短时交通量预测模型研究[D];长安大学;2012年
，本文编号：2463980
本文链接：https://www.wllwen.com/kejilunwen/daoluqiaoliang/2463980.html
