基于协同理论的城市路网状态特征信息提取与表达研究

项目名称： 基于协同理论的城市路网状态特征信息提取与表达研究

项目编号： No.51308021

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 建筑科学

项目作者： 于海洋

作者单位： 北京航空航天大学

项目金额： 23万元

中文摘要： 智能交通系统建设过程中的路网数据量呈几何级数增长，带来了信息超载和表达能力弱等问题，导致了系统的决策延迟或失误。因此，路网特征信息提取及表达技术研究已成为城市道路交通可持续发展的必然要求。本课题以协同理论为指导，通过对城市路网表征参量形式化定义建立有监督的信息保留准则，基于二维局部敏感辨别分析算法构建路网状态数据特征信息提取模型；在此基础上，对路网运行状态进行识别和分类，分别设计监督分类和非监督分类算法，建立基于特征参数提取的交通状态分类模型；利用路网状态划分定义路网状态的相，从而确定路网状态特征参量，实现二维路网状态演化过程的描述；利用城市路网状态特征信息一体化建模方法生成路网状态演化表达模型，进行状态特征信息展现。本课题能为城市交通管理和公众出行准确、快速地提供直观信息，为构建“和谐交通”提供有力的技术支撑。

中文关键词： 路网运行状态；特征信息提取；局部敏感判别分析；宏观基本图；深度学习

英文摘要： The amount of the road network data increases in a geometric progression during the construction of the intelligent transportation system, thus causes information overload and weakness of expression ability,which leads to the system decision delay or failure. Therefore, the characteristic information extraction and expression of city road network have become an inevitable requirement of the sustainable development of the city road traffic. In this research, under the guidance of the synergy theory, information retention criteria under supervision is established by defining the feature parameters of the road network, and the feature extraction model of traffic state is built based on two-dimensional local sensitive discriminant analysis algorithm. On the basis of the model, identification and classification of the road network running state are achieved, supervised classification and unsupervised classification algorithm is designed, and traffic state classification model based on feature extraction is built. The phases determining parameters of the network state are defined based on the division of network state, and the description of the two-dimensional network evolution state is realized. The network evolution state model is established by using the integrated modeling method of the city road network state ch

英文关键词： Urban Traffic Network；Feature Extraction；Locality Sensitive Discriminant Analysis；Macroscopic Fundamental Diagram；Deep Learning