Urban transportation systems play an important role in human activities. They are indispensible elements for the rapid and sustainable development of a city. Thus, research on traffic system has become a hotspot for years. Nevertheless, one problem that exists in the related studies is that it is always difficult or even infeasible to conduct experiments on real transportation system, due to its complexity nature and other reasons caused by economy, law, and ethic, etc. Artificial transportation systems (ATS) are based on the theories and methods of artificial societies. The main idea of ATS is to gain deep insight into vehicles movements and traffic evolution by extracting basic rules of individual vehicle (or local traffic behavior) and analyzing the complex phenomena emerging from interactions among individuals, further grow artificial transportation systems to be substitution for real transportation system. By conducting computational experiments in ATS, it is applicable to investigate transportation systems in a controllable, repeatable, observable way with little cost and risk. And we can explore the possible travel behaviors and evaluate the relevant decision-making in transportation systems. Thus, it is a valuable topic to study. In the concerned research, the design and analysis of computational experiments is a key matter. And this is what this dissertation focuses on. In this dissertation, the design and analysis of computational experiments in ATS is studied systematically on two aspects: theoretical foundation and specific cases. On the aspect of theoretical foundation, we summarize the developments of computational experiments, especially the state-of-the-art of computational experiments in ATS and the experimental platforms. After introducing the basic concepts, we propose the rules and general procedures of computational experiments, and present the design approaches of computational experiments that can be specifically used in ATS under different experimental scenaries. On the aspect of specific cases, we carry out experimental research to investigate the behaviors of main factors and the characteristics of the whole system. To be concrete, we conduct computational experiments respectively on three main factors (namely, population, network, and environment) of transportation system. Firstly, we design and analyze computational experiments on artificial population and travel behaviors which are related to factor “population”. We propo...
修改评论