【作者】 叶子超； 刘晓俐； 王晖；

【Author】 Ye Zichao;Liu Xiaoli;Wang Hui;College of Civil Engineering and Architecture,Zhejiang University;

【机构】 浙江大学建筑工程学院；

【摘要】 空间布局自动生成(Generation of Space Layouts,GSL)是建筑学研究的热点领域之一,目前按算法工具与理论基础归类,主要包括进化算法、模拟退火算法、数学规划、多智能体系统、基于物理的方法、基于图论的方法等,但各方法的分类研究未明确所对应阶段,造成了标准的不统一。本文将求解流程分为三个阶段,即预处理阶段、优化阶段与后处理阶段。其中核心的优化阶段依据采取策略的不同区别为分段优化与同时优化。分段优化策略通常分为拓扑优化与数值优化两个阶段。通过对最优化算法与启发式算法两类优化工具进行效用对比研究,分析其可扩能力、规模能力、全局能力、收敛能力。本文对各阶段的常用工具进行了初步评价,并为GSL系统的设计提出了一些参考原则。更多还原

【Abstract】 The automatic(GSL) is one of the popular areas of architecture research.According to algorithm tools and theoretical foundations,GSLs are currently classified into evolutionary algorithms,simulated annealing algorithms,mathematical planning,multi-agent systems,physics-based methods,and the method of graph theory and so on.But the classification research of each method does not clarify the corresponding stage,which has caused the inconsistency of the standard.This paper divides the solution process into three stages,namely the pre-processing stage,the optimization stage,and the post-processing stage.The core optimization stage is divided into two categories based on different strategies,namely multi-stages optimization and simultaneous optimization.Multi-stages optimization strategies usually include two stages,namely topology optimization and numerical optimization.Two types of optimization tools,optimization algorithms and heuristic algorithms,are analyzed to compare their expandability,scalability,global capabilities and convergence capabilities,which provides a reference for the selection of appropriate tools for a specific stage and some principles are put forward.更多还原