多智能体系统(Multi-Agent system,MAS)是一种进行复杂系统分析与模拟的强有力工具,尤其在社会科学领域得到了广泛的应用。本文提出了基于多智能体的居住区位选择模型(Agent-Based Model of Residential Location—ABMRL),将多智能体建模的方法应用于居民居住区位决策行为和地价动态变化的研究中,旨在探索与模拟居民在居住选择过程中的复杂空间决策行为,以及居民之间、居民与地理环境的相互作用而导致城市居住空间分异的演化过程。ABMRL模型由表征各类居民的多智能体层和表征地理环境的元胞自动层组成,对应人地关系中的两个基本要素——人类与自然环境。该模型认为居民迁居的动力源于内部的经济社会压力和外部的居住环境刺激。利用ABMRL模型模拟和验证了居住空间分异、圈层城市空间结构、城市绅士化等经典城市理论,并以广州市海珠区为实验区,模拟了该区域居民居住空间分异的演化过程和地价的动态变化。
在潜力模型基础上,提出了伪潜力指数,将其应用于定量描述城镇集聚能力的极化过程,并分析城镇规模和交通条件对集聚能力的影响。为了消除时间段上存在着常数的变量问题,通过面板数据分析(Panel Data Analy-sis)验证研究结果。将该方法应用于分析东莞各镇区1997-2005年城镇集聚能力的空间特征及其变化,探讨城市内部是否存在极化现象。结果表明,镇区集聚能力在空间上存在聚类;集聚能力的变化使极化现象在全市范围内逐步加强,但在经济核心地带内部出现了反极化现象。综合城镇规模对集聚能力的提高更加显著,其空间格局奠定了集聚能力的空间分布状况,从而决定了经济发展要素流动的方向,产生极化现象。选取净迁入人口和实际利用外资情况建立面板数据模型,估计结果证明潜力指数计算结果与人口和资金的吸引呈正相关。
This study proposes an integrated model based on small world network (SWN) and multi-agent system (MAS) for simulating epidemic spatiotemporal transmission. In this model, MAS represents the process of spatiotemporal interactions among individuals, and SWN describes the social relation network among agents. The model is composed of agent attribute definitions, agent movement rules, neighborhoods, construction of social relation network among agents and state transition rules. The construction of social relation network and agent state transition rules is essential for implementing the proposed model. The decay effects of infection "memory", distance and social relation between agents are introduced into the model, which are unavailable in traditional models. The proposed model is used to simulate the transmission process of flu in Guangzhou City based on the swarm software platform. The integration model has better performance than the traditional SEIR model and the pure MAS based epidemic model. This model has been applied to the simulation of the transmission of epidemics in real geographical environment. The simulation can provide useful information for the understanding, prediction and control of the transmission of epidemics.
