社会交互与出行行为的计算实验研究

CORC > 自动化研究所 > 中国科学院自动化研究所 > 毕业生 > 博士学位论文

题名	社会交互与出行行为的计算实验研究
作者	陈松航
学位类别	工学博士
答辩日期	2014-05-24
授予单位	中国科学院大学
授予地点	中国科学院自动化研究所
导师	王飞跃
关键词	人工交通系统出行行为社交网络社会交互社会化学习 artificial transportation systems travel behaviors social networks social interaction social learning
其他题名	A Study of Social Interaction and Travel Behaviors using Computational Experiments
学位专业	控制理论与控制工程
中文摘要	深入地了解人们的出行行为，一方面有助于在实际中改进对出行需求的分析和预测，从而促进合理的出行需求管理，缓解日益严重的交通拥堵等问题；另一方面，有助于在计算机上构建微观的、可靠的交通系统模型，从而支撑各种在实际中无法开展的实验，寻找解决实际交通问题的有效方法。人是社会性动物，日常生活中不断地与他人进行着交互。随着社会经济和科学技术的发展，社交网站等新型的信息交互媒介不断出现，人们的社会交互方式日益多样，出行行为也受到越来越多社会因素的影响。为此，研究社会交互对个体出行行为的影响，促进对人们出行行为的认识，具有重要的理论价值和实际意义。本文在人工交通系统理论的基础上，结合社交网络，行为学，交通学等相关的理论方法，研究了人工人口的社交网络生成方法和基于社会化学习的出行行为模型，在此基础上设计和开展了社会交互影响个体出行行为的计算实验。论文的主要工作包括以下几个方面：首先，对相关研究工作进行了梳理，说明了人工人口的概念和生成方法，概述了三类可用于人工人口的社交网络生成研究，并介绍了国内外的出行行为研究现状。其次，论述了基于活动和代理的出行行为建模，详细介绍了已有的工作基础TransWorld——基于人工交通系统的计算实验平台，及其采用的基于代理的人工人口系统和基于活动的出行行为建模。再次，提出了一种基于人工交通系统的社交网络生成方法，所生成的社交网络具有与实际调查相符的拓扑结构和空间特性。在此基础上，进一步将社会化学习引入到代理的出行行为选择中，提出了三种考虑社会交互的出行行为模型，并定义了三种考察指标，分别反映人工人口经过学习后做出最佳选择、习惯性选择以及流行性选择的程度。最后，设计和开展了社会交互影响个体出行行为的计算实验研究，选取中关村局部区域作为实验的参考场景，研究了人工人口的休闲类活动场所选择，以及社会交互因素和社交网络对个体出行行为的影响，分析了人工交通系统中涌现出的人工人口出行行为变化。
英文摘要	A deep understanding of people’s travel behaviors contributes to actual travel demand prediction and analysis. Benefit from this, travel demand management can be improved to alleviate worsening traffic problems, like congestion and pollution. On the other hand, a deep understanding of people’s travel behaviors also contributes to modeling microscopic and reliable transportation simulation models on computers. With the reliable models, the experiments that cannot be conducted in reality can be done on computers to search effective and satisfiable solution of real traffic problems. Human beings are social animals. With the development of social economy and technology, new interaction media like online social networks is emerging, and the forms of people’s social interaction become increasingly diverse, so that their travel behaviors are influenced by more and more social factors. Therefore, a study on how social interaction affects individual travel behaviors, which can improve the understanding of people’s travel behaviors, is of great important significance theoretically and practically. This dissertation is based on the artificial transportation systems (ATS), combined with social network, behavior science, traffic science, and other related theories. Social network generation of artificial population and travel behavior model based on social learning are implemented at first. Then, computational experiments on the influence of social interaction on individual travel behaviors are designed and conducted. Main works of this dissertation are as follows. Firstly, related works are reviewed. The concept of artificial population as well as its generation methods is introduced. Three kinds of social network generation are described. Some studies on travel behavriors are demonstrated. Secondly, agent-based and activity-based travel behavior modelling is addressed. TransWorld, a computational experiment platform based on ATS, is introduced in detail as well as its agent-based artificial population and activity-based travel behavior modeling. Thirdly, a novel method based on ATS is proposed to generate social networks of artificial population. The generated social networks have similar topological and spatial characteristics to real social networks. Further, social learning is introduced into agents’ decision making on travel behaviors. Three travel behavior models, which take social interaction into account, are presented. Three measure indexes are ...
语种	中文
其他标识符	201118014628005
内容类型	学位论文
源URL	[http://ir.ia.ac.cn/handle/173211/6593]
专题	毕业生_博士学位论文
推荐引用方式 GB/T 7714	陈松航. 社会交互与出行行为的计算实验研究[D]. 中国科学院自动化研究所. 中国科学院大学. 2014.