整合大数据挖掘与路网脆弱性评估的疏散交通管理研究

项目名称： 整合大数据挖掘与路网脆弱性评估的疏散交通管理研究

项目编号： No.51478151

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 建筑环境与结构工程学科

项目作者： 安实

作者单位： 哈尔滨工业大学

项目金额： 65万元

中文摘要： 道路交通应急疏散管理的核心理论瓶颈在于灾害条件下疏散者行为数据调查难度较大，导致相应疏散交通需求预测精度较低。同时，基于最优化思想的疏散交通管理策略优化，由于建立在诸多理想假设之上，致使其实际应用效果较差。本项目针对灾害背景下的应急疏散交通管理问题进行研究，首先基于大数据集挖掘疏散者灾害前后的出行行为，从而建立疏散个体活动-出行链模型，准确预测疏散交通需求；以此为基础，识别出路网中脆弱而关键的疏散需求节点和疏散通道，进而针对二者分别构建节点层和通道层交通管理优化模型；最后面向疏散交通管理者决策的实际需要，研发疏散行为挖掘与交通管理策略仿真及决策支持原型系统，并以美国艾琳飓风疏散的实际情景验证研究成果的合理性和实用性。本研究整合了大数据挖掘技术、基于脆弱性评估的疏散交通管理优化理论与基于活动的出行需求建模理论，这种三位一体的研究方法，相对于传统的疏散交通管理研究构成了重要创新。

中文关键词： 应急；区域疏散；交通管理；脆弱性；数据挖掘

英文摘要： The main theoretical bottleneck on road traffic management during emergency evacuation is that evacuees' behavior data survey is too difficult to conduct, which directly leads to inaccurate evacuation traffic demand forecast results. Meanwhile, the optimization framework to obtain evacuation traffic management strategies contains too many ideal hypothesis, the desirable optimum strategies usually show little effect in the real application. The program studies the issues on emergency evacuation traffic management under disaster. Firstly, we study evacuee's travel behavior before and after the disaster based on big data mining, and then construct an activity-travel model for individual evacuee to accurately predict the evacuation demand. Using as a base, the vulnerable demand nodes and critical evacuation paths are identified, and traffic management optimization models are built on node layer and path layer respectively. Finally, according to the actual requirement of evacuation management decision-maker, a prototype decision-making support system integrated with evacuation behavior mining and traffic management strategies simulation is developed. The reasonability and practicality are verified in the evacuation case of 'Irene' hurricane in United States. In this program, three theories including big data mining technique, optimization theory of evacuation traffic management based on vulnerability assessment, and activity based travel demand modelling theroy are integrated to form an organic whole method. This 'trinity' method is an important innovation with respect to traditional evacuation traffic management research.

英文关键词： Emergency;Regional Evacuation;Traffic Management;Vulnerability;Data Mining