Construction of volume generation design platform system driven by deep learning
Year: 2025
Type: article
Abstract: Based on the latest research progress of artificial intelligence technology, this study tries to introduce deep learning to drive the volume generation design of architectural conceptual schemes. The purpose of this study is to realize the automation of building volume generation design by intelligent technology. Three technical realization paths of building volume generation design based on image translation, building volume deepening generation design based on image completion and building volume deepening generation design based on style transfer are focused on. The realization principle of each technical path is discussed in detail, and the corresponding deep neural network driver program is developed. The research systematically analyzes the algorithm principle, program framework, condition parameters and control parameters of each driver; The training and use methods of each driver are described respectively. And summarizes the design flow of function realization under each technical path. The research also combined with specific design cases to carry out the application and verification of each technical path and platform system, through the systematic analysis of the design process and generated results of each path, and finally concluded that the design method and platform system proposed in this study have better design effect and design efficiency. It provides useful enlightenment and reference for the research of building intelligent design, computational design theory and technology tool development and other related fields.
Authors: Guiqing Lv, Yanxin Zhao, Jing Liu, Yiran Luo
Language: English
Subfield: Architecture
Field: Engineering
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4408766949