基于生成对抗网络的复合功能体系计算性设计——以职业技术学院校园平面生成为例

Computational Design of Multi-functional System Based on Generative Adversarial Networks Taking the Layout Generation of Vocational and Technical College as an Example

选取职业技术学院作为研究对象,建立职校平面的数据集,并基于生成对抗网络进行机器学习和模拟。通过机器学习对平面图的逐步生成和人工优化的工作方式,本研究分为3个阶段,其中包含区域划分、位置确定、形态生成3个步骤。机器学习模型可以快速生成包含主要空间元素互动组合的研究型方案,进而帮助建筑师探讨职校设计的规律性和神经网络的外延性。

This paper selects vocational and technical colleges as the research object and established a data set of vocational and technical college plans,for conducting machine learning and simulation based on Generative Adversarial Networks.The working mode of this research is the gradual generation and manual optimization of architectural plans by machine learning,and this research process is divided into three stages including region division,location determination and form generation.The machine learning model can rapidly generate a series of research-based schemes which contain the interaction combination of major spatial elements,and help architects to explore the regularity of vocational and technical college design as well as the extension of neural network.