摘要
人工智能技术的发展为设计师提供了新的设计工具,将深度学习技术引入到城市设计中,提出基于pix2pix算法模型的建筑形态及其组合生成式设计方法,优化设计流程,提升效率。在城市设计中,建筑形态及其组合是城市空间要素之一,影响着城市空间结构。图底理论将空间结构与平面形体几何学建立联系,建筑物与空间互为图形和背景,以此为基础,将建筑平面轮廓和道路作为图形,空间作为背景,生成二维平面图象,以独立型、网格型、围合型3种建筑形态及其组合为例,使用pix2pix算法模型学习到这3种类型的参数信息,并应用于新的设计基地中,可以在1s内生成建筑群平面模型,为设计师提供更多可行的参考。
The development of artificial intelligence technologies provides designers with new design tools.Introducing deep learning technology into urban design,a generative design method for architectural form and combination based on the pix2pix algorithm model was proposed to optimize the design process and improve efficiency.In urban design,architectural form and combination are one of the urban spatial elements and influence the urban spatial structure.The figure⁃ground theory establishes the connection between spatial structure and planar form geometry,and buildings and spaces are each other as graphics and backgrounds.The building plan outline and roads were used as graphics and space was used as background to generate 2D plan images.Taking the three architectural form and combination of“independent type”,“grid type”and“enclosure type”as examples,and using the pix2pix algorithm model to learn the parameter information of these three types and create a generator.Then,the generator was applied to a new design cite and generate the building group plan model within 1s,providing more feasible reference for designers.
作者
刘德利
王科奇
LIU Deli;WANG Keqi(School of Architecture and Urban Planing,Jilin Jianzhu University,Changchun 130118,China)
出处
《建筑科学》
CSCD
北大核心
2022年第10期260-267,286,共9页
Building Science
关键词
深度学习
城市设计
图底理论
建筑形态及其组合
生成式设计
deep learning
urban design
figure⁃ground theory
architectural form and combination
generative design
作者简介
刘德利(1997⁃),男,在读硕士研究生;通讯作者:王科奇(1971⁃),男,博士,教授,联系方式:wkq0431@126.com。
