基于自主体(agent)的单中心城市化动态模拟

构建了基于自主体(agent)的单中心城市化模拟模型,实现了一个人工可控的计算实验环境。围绕人口流动、扩大内需、集聚效应等城市化进程的影响因素以及城市化与工业化关系展开模拟分析和讨论。研究表明:(1)城乡消费倾向扩大有助于城市与农村地区收入的提高,也有助于提高城市化水平;(2)随着集聚过程的负效应增大,城市化水平相应下降,甚至不可持续;(3)城市化与工业化具有较强的相关性,经济结构比值随着城市化水平上升而不断提高,而且工业化进程要快于城市化过程。本文认为,基于自主体建模为研究和揭示我国城市化过程的动态规律和驱动机制提供了一条新的途径。我国城市化研究需要结合转型时期的基本国情,探索演绎分析、计算实验以及实证研究相结合的技术路线。

Dynamic simulation of impact factors on mono-centric urbanization by using agent-based modeling

In this paper, the UrbanSwarm, a computational environment, is developed for a better understanding of urbanization by agent-based modeling. UrbanSwarm contains multiple economic interactions between urban and rural agents that are crucial in urbanization process and mechanism because the basic force driving the urbanization is inherently microscopic. This paper gives a detailed discussion on the influencing factors such as migration policy, domestic demand, and negative effect of agglomeration on urbanization process in mono-centric scenario. (1) The segmentation policy not only limited the urban-rural migration but enlarged urban-rural income disparity as well. It simulates the household registration system (hukou) before 1978, which confined most Chinese citizens to their places of birth. (2) Raising urban-rural propensity to consumer contributes to both urban-rural income level and urbanization level. Meanwhile, it can narrow the urban-rural gap. (3) With the increase of the negative effect of agglomeration on urbanization, urbanization level drops gradually. The urbanization process loses its sustainability if the negative effect of agglomeration on urbanization exceeds some thresholds in mono-centric pattern. (4) The industrialization and urbanization level has a close relationship if the technology and preference are never changed. The process of industrialization is faster than the process of urbanization in simulation. This parallel processing agent-based approach has various advantages over existing economic approaches such as neoclassical model and equilibrium analysis. UrbanSwarm can be used for both simulation and computation purposes. MAS provides an exploratory platform to test hypotheses behind the space-time dynamics as well as to experiment with 'what-if' games within complex urbanization process, i.e. using the computer as an artificial laboratory for the study of urban and regional systems. Besides, it can be used to achieve some computational intensive tasks through the collective work of individual agents such as rural and urban households, local government, firms and so on. Future work should focus on removing the limitations of reactive agents, implementing the use of cognitive or deliberate agents and achieving their full integration into GIS environment.