【作者】 张鸿辉；

【导师】 曾永年；

【作者基本信息】 中南大学 ， 地图制图学与地理信息工程， 2011， 博士

【摘要】 城市空间系统是一个复杂系统,其空间演化以大量具有能动性、适应性的微观行为主体(企业、居民、农民以及各类组织等)以及行为主体与环境的非线性相互作用为特征,产生出非连续的城市空间决策行为。传统的城市规划空间决策模型因其静态特征,无法反映出大量微观行为主体在城市规划空间决策互动行为中的时间维累积和空间维聚集过程,较难有效表达城市规划参与者的空间决策行为,规划结果的现势性、可接受程度均受到了一定限制。本文面向城市规划的实际需求,尝试结合多智能体系统、地理信息科学、计算机科学以及智能科学的相关理论和技术,构建可以清晰表达城市规划参与行为主体决策时空显性特征的城市规划空间决策模型,并基于所构建模型,开展了长沙市城市规划空间决策模拟与分析应用研究,辅助认识和解决城市规划过程中所遇到的相关非结构化问题,以期提高城市规划空间决策的科学性与合理性,为促进城市的可持续发展决策提供先进的技术方法与参考借鉴。主要的研究结果如下：(1)设计了基于多智能体交互的城市规划空间决策行为和规则体系,构建了以多智能体系统为核心的城市规划空间决策系列模型。较基于元胞自动机的城市规划空间决策模型,在城市空间系统演化成因解释方面的能力大大提高,同时还可弥补传统的基于GIS的城市规划空间决策模型在过程模拟能力方面的不足,加强其表现力,有助于人们对城市空间系统演化表象之下的作用机制和交互过程的理解。(2)以多智能体系统理论为基础,建立城市土地资源时间和空间配置规则,构建了动态且能描述影响城市土地扩张的智能体间互动关系的城市土地扩张模型,模型对于解释城市土地扩张的成因、理解智能体行为对城市土地扩张过程的影响是合适的,且具有较高的模拟精度,能够为政府和城市规划者制定用地政策提供辅助决策支持。(3)在多目标约束下,结合城市土地利用空间优化配置问题的实际,集成多智能体系统与启发式算法构建了多智能体进化城市土地利用空间优化配置模型,该模型不仅能获取合理、可行的城市土地利用优化配置结果,并且具有良好的运行效率和目标优化能力。(4)在多智能体系统支持下,构建了一种基于信息动态反馈机制的多智能体交互式空间选址模型,使参与城市规划的智能体之间频繁的交流各自所拥有的知识与信息,最终达成一致决策意见,形成对选址问题的解决方案。该模型所设计的信息动态反馈机制能够有效的表达智能体的交流与学习过程,也使多智能体交互式选址模型优于传统选址模型中常用的空间叠加分析方法,从而为城市建设项目用地选址提供行之有效的解决方案。(5)以中部地区典型城市、长株潭城市群“两型社会”改革试验区核心区域—长沙市市区为例,基于所构建模型,开展了长沙市城市土地扩张、多目标城市土地利用空间优化配置、建设项目选址应用研究,模拟和分析了长沙市城市规划空间决策问题,为该区域的城市规划和管理提供了理论依据和决策支持。多智能体城市规划空间决策模型还处于理论探讨阶段,留下了巨大的后续研究空间,其理论深度、模拟精度、可推广性有待进一步探讨。更多还原

【Abstract】 Urban spatial system is a complex system. Its spatial evolution is characterized by j a number of microscopic behavior agents (enterprises, residents, farmers and various organizations) with activity and adaptability and nonlinear interaction between agents and the environment,which results in non-continuous urban spatial decision-making behaviors. Due to static characteristics, traditional model of urban planning spatial decision-making can’t reflect the accumulation process of time dimension and aggregation process of spatial dimension in urban planning spatial decision-making interaction by large number of microscopic agents. Therefore, the traditional way is difficult to embody the essence of public participation in urban planning decision-making, and actuality and acceptability of urban planning are subject to a certain restriction. Based on actual needs of urban planning and management, this thesis tries to combine multi-agent system theories and technologies related to geographical information science, computer science and intelligence science, and build urban planning spatial decision-making model which can clearly expresses spatio-temporal explicit characteristics of urban planning decision-making by agents participating. On basis of the model, it carries out simulation of urban planning spatial decision-making for Changsha city, and assists understanding relevant non-structured problems arising from urban planning management, so as to improve the scientificity and reasonability of urban planning spatial decision-making and provide reference to sustainable urban development in China. Main conclusion of the research includes:(1) This study has designed urban planning spatial decision-making behaviors and rule system based on interaction among multi-agents and built series of urban planning spatial decision-making models focusing on multi-agent system. Compared with urban planning spatial decision-making model based on cellular automaton and traditional urban planning spatial decision model based on GIS, multi-agent urban planning spatial decision-making model improves greatly the ability of interpretation of the reasons for urban spatial system evolution and simulation of evolution process of urban spatial system, respectively, which enhances its capacity of representation of urban spatial evolution and helps people to understand the mechanism of action and interaction process under the presentation of urban spatial evolution.(2) Based on multi-agent system theory, this study has established spatio-temporal allocation rule for urban land resources and built dynamic urban land expansion model which is able to describe interaction relationship among agents affecting urban land expansion. The model is suitable for interpretation of reasons for urban land expansion and understanding of impact of agents’behaviors on urban land expansion process, with higher simulation accuracy, so it can provide auxiliary decision-making support on formulation of land use policies for government and urban planners.(3) Restricted by multiple objectives and based on actual problems in spatial optimal allocation of urban land use, this study has built spatial optimal allocation model of urban land use integrating multi-agent system and heuristic algorithm. This model not only can obtain reasonable and feasible optimal allocation result of urban land use, but also has good operation efficiency and objective optimization capacity.(4) With support of multi-agent system, this study has built multi-agent interactive spatial site selection model on the basis of dynamic information feedback mechanism, enabling agents participating in urban planning to communicate their own knowledge and information frequently and reach consensus on decision making ultimately, so as to get the solutions of site selection problems. The dynamic information feedback mechanism designed in the model can effectively express the communication and learning process of agents, and enables multi-agent interactive spatial site selection model to be better than spatial overlay analysis commonly used in traditional site selection model, so as to make the model provide effective solutions to site selection for construction project. (5) Taking Changsha, the typical city in central China and the core area of Changsha, Zhuzhou, Xiangttan city group, which is the national comprehensive reforms test areas of building resource-saving and environment-friendly society, as an example, urban land expansion, multi-objective spatial optimal allocation of urban land use, construction project site selection of changsha city was simulated and analyzed based on developed urban planning spatial decision-making model, so as to provide theoretical basis and decision-making support for urban planning management of this region.Multi-agent urban planning spatial decision-making model is still in the phase of theoretical study, leaving a huge space for follow-up research, and its theoretical depth, simulation accuracy and applicability still need further exploration.更多还原