【摘要】 智能体行为模拟技术在空间使用分析与测度中被广泛应用。文章以上海市北外滩滨水公共空间为例，搜集其2021年秋季用户空间行为数据，还原最主要空间活动需求与微观选择偏好，并采用聚类与回归算法构建智能体行为规则，在平面空间中进行空间使用模拟。结果显示：宽阔空间与集中式设施布局引导活动在空间与时间尺度上呈点状聚集，造成局部活动干扰与空置空间较多的不均衡现象；近水空间设施分散化引导活动呈离散式分布且局部仍然存在干扰；支路空间则利用率较低。研究依托模拟结果对各类空间的设施离散布置方式、功能多样性、空间尺度、空间连接提出优化，以改善活动品质，体现行为模拟在空间优化中的具体流程。更多还原

【Abstract】 Agent-based simulation technology is widely applied in spatial usage analysis and measurement. This study focuses on the waterfront public space in North Bund, Shanghai, by collecting user spatial behavior data during autumn 2021. It reconstructs the primary spatial activity demands and micro-choice preferences of users and employs clustering and regression algorithms to model agent-based behavioral rules.Spatial usage simulations are then conducted in a planar space. The results indicate that wide-open spaces and centralized facility layouts lead to clustered activities across spatial and temporal scales, causing localized activity interference and underutilized spaces. Dispersed facilities in waterfront areas result in scattered activity distributions, yet localized interference persists, while branch spaces remain underused. Based on these simulation outcomes, the study proposes optimizations for facility dispersion, functional diversity, spatial scale, and connectivity to enhance activity quality, illustrating the specific process of how behavioral simulations can inform spatial design improvements.更多还原