【摘要】 大型公共建筑的寻路效能是其建设前后重要的关注点，传统上常用现场调研或实验室实验方法加以评估，但受限于能参与的被试人数，无法适应越来越多的大规模复杂空间。计算机人流模拟方法应运而生。目前，它普遍遵循“采集—建模—模拟”的范式，即经历1个从一般建筑中采集一般人群行为数据，到归纳构建数学行为模型，再对目标建筑中的目标人群应用模型来迭代生成路径的过程。该范式存在数学模型构建前后，所面对的建筑空间不一致、人群不一致、归纳环节抑制复现行为多样性等三大问题。本论文提出了“采集即模拟”的思想，充分利用互联网远程虚拟实验采集技术，依托其对大量真人寻路路径数据的快速采集能力，配合时空挪移复现算法，实现了采集建筑与模拟建筑、采集人群与模拟人群的完全一致，并完整保留了人群行为多样性，为建筑空间寻路效能的前、后评估提供了新范式。该方法在上海浦东国际机场卫星厅的后评估项目中予以应用，在72 h的在线实验开放期内完成了13 712人次数据的采集，圆满完成了对整个空间的模拟评估任务。更多还原

【Abstract】 Wayfinding efficiency is one of the key issues in pre-occupancy or post-occupancy evaluation for large public buildings. It traditionally was conducted through field observations or laboratory experiments. However, the limited number of subjects prevents these methods from applying to large-scale and complex spaces. Thus, computer simulation of pedestrians was created. It has a “sampling-modeling-simulating” paradigm, an iteration-based path generation process through the collection of crowd behavior data from general buildings, establishing mathematical behavior model to applying the model to target crowd in the target building. This still faces three bottle-necks of inconsistent building space and crowd before and after the modeling, and data induction inhibiting the diversity of pedestrian behavior. This study proposed an approach featuring the idea of “sampling as simulating”. It makes full use of Web-based virtual experiment technologies, uses its capability in rapidly collecting a large number of real human users’ wayfinding data, and adopts the time-shift and replication algorithms to realize the consistency between the sampled building and simulated building, sampled crowd and simulated crowd. This has kept the diversity of subjects’ behavior, and provided new paradigm for the pre-and post-evaluation for building space wayfinding efficiency. Such a new pedestrian simulation approach was validated in a post-occupancy-evaluation project of Pudong Airport Satellite Hall through 13 712 sets of data collection within a 72-hour window time.更多还原