Study On Microscopic Simulation Model Of Pedestrian Behavior In Walking Facilities

With the progress of human civilization, many long-distance multi-mode of travel as well as various group activities have become increasingly frequent. The pedestrian infrastructures such as transportation hubs and sports stadium highlight the significant role in social life. Based on consideration of pedestrian safety, efficiency and comfort, the microscopic pedestrian simulation aim to evaluate and optimize the construction and operation of walking facilities, which has become a hot research topic. This dissertation focuses on simulation model of pedestrian behavior in large-scale walking facilities. The architecture of this dissertation is: (1) Analysis of simulation research framework; (2) Path planning algorithm; (3) Walk simulation model; (4) Facility interactive model; (5) Simulation software development and application.First, this dissertation reviews existing literature on pedestrian behavior simulation modeling. It systematically summarizes the main simulation models and analyses the basic principles, assumptions and the core algorithm of force-based model, the cellular automata simulation model and the discrete choice model. Moreover, this dissertation emphasizes the critical limitation of force-based model: the linear synthesis mechanism for effect factors. Based on the features of large-scale multi-layers walking facility such as complex spatial structure and a variety of pedestrian behaviors, a continuous spatial mechanism includes basic walk zone and transport facility is proposed. The pedestrian's decision-making behavior is divided into macro, meso and micro three levels. The overall logic flow of pedestrian behavior and research framework is extracted. Moreover, their applicability in the normal operating and emergency evacuation is analyzed. The walk simulation model, path planning algorithm and facility interactive model are focused.Through the analysis on single-step walk movement, a new description mechanism is proposed including direction changing and velocity adjusting. Analyzed pedestrian's actual decision-making process, a bran-new walk decision-making model based on neighborhood impact principle is designed. Its core principle is: pedestrian's decision is made according to his inner motivations and all information from his neighborhood. This model includes the description method, the analyse algorithm for neighborhood and the subjective utility calculation method for impact factors. This dissertation also designs three path planning algorithms based on diffirent elements, which are the Shortest Path Algorithm, the Natural Path Algorithm and the Discrete Potential Field-based Planning Algorithm. Their logic flow, advantages and applicability are introduced in detail. In order to modeling multi-layer space, this dissertation proposes an interactive mechanism and modeling method between basic walk zone and transport facility. For instance, a stairs model and escalator model are discussed.Based on the above-mentioned research results, this dissertation finally develops a microscopic pedestrian simulation software prototype named RealWalker using Visual C++ 2008. The RealWalker can simulate the large-scale multi-layer walking facilities and support a complete workflow includes the scene construction, simulation execution, output and analysis. Finally, the simulation experiments verified the validity of each model, the ability of mutual cooperation of multi-models and the applicability in the normal operating and emergency evacuation.