多智能体模拟在规划师知识构建及空间规划决策支持中的应用--以日本地方城市老年人日护理中心空间战略规划为例

Application of Agent-based Simulation in the Process of Knowledge Formation of Planners and Spatial Planning Support and Decision: A Case Study of Spatial Strategic Planning of Daycare Center for the Elderly in Japanese Local Cities

面向城市问题的复杂性和动态性,大量模拟仿真、定量分析等手段被广泛应用于城乡规划方案的预测与评价中,而其相应的预测评价结果也往往被认为有助于规划方案的调整与决策。在此过程中,仿真模型的构建逐渐从宏观向微观方向发展,对研究对象的分析也逐渐体现出从群体性描述向个体行为特征、机理的刻画转变。但是这其中却鲜有学者探讨基于微观个体行为的仿真模拟究竟在规划方案的生成与决策过程中发挥怎样的作用,对规划师的知识构建是否能够有效地达到规划决策支持的目的。本研究从这一科学问题出发,基于一个面向养老日护理中心空间分布模拟的多智能体模型(Agent-Based Model,ABM)的构建与应用,探讨多智能体模拟仿真在空间规划实践中的应用可能及其构建机制。研究结果表明,多智能体模拟用于规划实践首先必须具有可信的、完整的数据支持;要达到辅助规划师建立对规划问题的科学认识,模型构建则应该在规划工作的最初开始介入,并让规划师参与其中;同时,多智能体模型的构建必须基于面向规划问题的现实决策过程。总体而言,多智能体模型可用于规划决策支持,但是模拟结果却不能代替规划方案。

A Large amount of simulation and quantitative analysis have been applied in the prediction and estimation of urban planning in response to the complexity and dynamism of urban problems, and the results of prediction and estimation are usually considered helpful for the adjustment and decision of planning. During this process, transformation in the design of simulation model is emerging, which are from macro to micro and from description of group to analysis of character and mechanism of individual behaviors. However, researches are rare on how the simulation based on individual behavior works during the process of generation and decision of planning proposal and whether it can effectively support the planning decision in the knowledge formation of planners. To explore above scientific question, this research is trying to study the application and design mechanism of agent-based simulation in spatial planning based on an agent-based model, which deal with the spatial distribution simulation of daycare center for the elderly. The research results indicate that reliable and complete data is of first importance to applying agent-based simulation in planning practice, and the model should be designed at the beginning of planning which should involve planners＇ work to assist planners to understand planning problems scientifically. Meanwhile, agentbased model must be aimed at the realistic decision process of planning problems. As a conclusion, it is possible to apply agent-based model to support the planning decision, but the planning proposal cannot be replaced by simulation results.