基于多智能体与最短路径算法的建筑空间布局初探--以住区生成设计为例  

Exploration on Architectural Space Layout Based on Multi-agent and Dijkstra Algorithm--A Case of the Residence Generative Design

作　　者：张柏洲 李飚[1] Zhang Baizhou;Li Biao(School of Architecture,Southeast University,Nanjing Jiangsu 210096,China)

机构地区：[1]东南大学建筑学院,江苏南京210096

出　　处：《城市建筑》2020年第27期7-10,20,共5页Urbanism and Architecture

基　　金：自然科学基金面上项目(51978139);国家自然科学重点基金项目(51538006)。

摘　　要：随着数字技术在建筑领域应用的不断深入,生成设计方法逐渐展示出其潜能。生成设计以建筑原型提炼为起点,以程序算法模型构建为核心,实现数字驱动下的设计生成,是建筑设计技术理性与艺术灵感的有机结合。本文通过生成设计的手段,探索住区规划的自组织生成方法,基于多智能体系统,结合Voronoi剖分、Dijkstra寻径等算法,解决住区规划中地块划分、建筑布局、道路生成等设计问题,为建筑空间布局的程序算法应用提供参考。With the deepening application of digital technology in the field of architecture, generative design gradually shows its potential. Generative design starts with the abstraction of architectural prototype, takes the construction of algorithm model as the core, and realizes digital-driven design. It is an organic combination of technical rationality and artistic inspiration. This paper explores the method of selforganizing generation of residential planning by means of generative design. Based on multi-agent system, Voronoi partition and Dijkstra algorithm, etc., it solves the design problems of plot division, building layout and road generation in residential planning, and provides a reference for the application of program algorithm of architectural spatial layout.

关 键 词：生成设计 住区规划 多智能体 地块划分 Dijkstra寻径算法 

分 类 号：TU201[建筑科学—建筑设计及理论]