高速公路交通流混沌特性分析及其在流量预测中的应用

发布时间：2018-05-23 08:01

本文选题：交通流 + 混沌　；参考：《重庆大学》2014年硕士论文

【摘要】：对高速公路交通流量实时准确的预测是制定合理的交通管理调控方案，有效地诱导出行者合理选择行驶道路的重要前提。高速公路系统是个复杂的开放系统，其交通流表现出周期性和不确定性共存的特性，而这恰恰正是以非线性和自相似性为特征的混沌系统的具体体现。因此，本文引入混沌理论对高速公路交通流特性进行研究，建立混沌预测模型，提高高速公路交通流量预测精度，进而提升对高速公路的诱导调控能力。本文在分析高速公路交通流混沌特性的基础上，通过重构相空间，建立高速公路短时流量预测模型对高速公路交通流量进行预测。本文的具体研究工作总结如下： ①高速公路交通流混沌特性的识别及其关联性的分析。首先利用最大Lyapunov指数法和关联维数法分别验证了高速公路交通流量、速度和占有率时间序列的混沌特性；然后分析了高速公路交通流各参数的混沌关联性，为第五章建立多参数预测模型的研究提供前提支撑。 ②针对很多文献都一直规避的基于最大Lyapunov指数的混沌预测会出现两个预测值的问题引入马尔科夫链改进最大Lyapunov指数的混沌预测方法。改进的方法将时间序列的斜率作为状态变量，并根据马尔科夫链建立状态转移矩阵，进而判定预测值演化方向。 ③高速公路交通流系统是个复杂的混沌系统，仅用流量时间序列重构相空间不一定能勾勒出系统完整的混沌特性。针对这种情况，本文首先根据Bayes估计理论将高速公路速度和占有率时间序列融合到一个新的相空间，该相空间包涵了速度和占有率的混沌特性；然后以流量相空间的相点作为基础重构分量，辅以新融合相空间的相点作为重构变量，，通过条件熵扩维的方法将两个相空间进行融合重构，进而为建立多参数预测模型提供简洁充分的信息。本文以渝武高速公路交通流量数据分别对建立的单参数预测模型和多参数预测模型进行了验证。结果表明：本文改进的最大Lyapunov指数预测法具有较高的的有效性和可行性；本文改进的多参数相空间重构方法能够更好地勾勒出系统的混沌特性，并且基于其建立的多参数预测模型具有较高的预测精度。
[Abstract]:The real-time and accurate prediction of expressway traffic flow is an important prerequisite to formulate reasonable traffic management and control scheme and to effectively induce travelers to choose a reasonable road. Expressway system is a complex open system, and its traffic flow is characterized by periodicity and uncertainty, which is exactly the embodiment of chaotic system characterized by nonlinearity and self-similarity. Therefore, this paper introduces chaos theory to study the characteristics of freeway traffic flow, establishes chaotic forecasting model, improves the precision of expressway traffic flow prediction, and then improves the ability of inducing and regulating the expressway. On the basis of analyzing the chaotic characteristics of freeway traffic flow, this paper establishes a short-time expressway traffic flow forecasting model by reconstructing the phase space to predict the expressway traffic flow. The specific research work of this paper is summarized as follows: 1. Identification of chaotic characteristics of freeway traffic flow and analysis of its correlation. Firstly, the chaos characteristics of freeway traffic flow, velocity and occupancy time series are verified by maximum Lyapunov exponent method and correlation dimension method, and then the chaotic correlation of various parameters of freeway traffic flow is analyzed. In the fifth chapter, it provides the premise support for the research of the multi-parameter prediction model. (2) to solve the problem that the chaos prediction based on the maximum Lyapunov exponent has been evaded by many literatures, the Markov chain is introduced to improve the chaos prediction method of the maximum Lyapunov exponent. The improved method takes the slope of the time series as the state variable and establishes the state transition matrix according to Markov chain and then determines the evolution direction of the predicted value. 3 the freeway traffic flow system is a complex chaotic system, only the reconstruction of phase space with flow time series can not outline the complete chaotic characteristics of the system. In this paper, the freeway velocity and occupancy time series are fused into a new phase space according to the Bayes estimation theory. The phase space contains the chaotic characteristics of velocity and occupancy rate. Then the phase points of the flow phase space are taken as the basic reconstruction components, and the phase points of the new fusion phase space are used as the reconstruction variables, and the two phase spaces are fused and reconstructed by the method of conditional entropy expansion. Furthermore, it provides concise and sufficient information for the establishment of multi-parameter prediction model. Based on the traffic flow data of Yuwu Expressway, this paper verifies the single parameter prediction model and the multi parameter prediction model. The results show that the improved maximum Lyapunov exponent prediction method is more effective and feasible, and the improved multi-parameter phase space reconstruction method can better describe the chaotic characteristics of the system. And the multi-parameter prediction model based on it has high prediction accuracy.
【学位授予单位】：重庆大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U491.1

【参考文献】