基于大语言模型的建筑平面图结构生成

Journal: Building Development DOI: 10.12238/bd.v9i3.4386

张百慧

南京大学建筑与城市规划学院

Abstract

随着大语言模型(Large Language Model,LLM)在智能推理与自然语言理解方面能力的持续突破,建筑平面图的数字化生成正迎来新的范式转变。本文提出一种基于大语言模型驱动的平面图结构生成方法,旨在模拟建筑师的思维过程,实现从用户自然语言需求到建筑空间图结构的自动化转换。通过引入思维链方法与建筑知识注入机制,大语言模型能够从非结构化文本中学习提取房间功能、空间关系与设计逻辑,并生成结构化的功能拓扑图。本研究不仅拓展了建筑生成设计的技术边界,也为多轮交互式设计工具的开发提供了新思路。

Keywords

大语言模型;建筑平面生成;图结构;思维链

References

[1] Lobos D,Donath D.The problem of space layout in archit ecture:A survey and reflections[J].Arquitetura Revista, 2010, 6(2):136-161.
[2] Parente J,Rodrigues E,Rangel B,et al.Integration of co nvolutional and adversarial networks into building design: A review[J].Journal of Building Engineering,2023,76:107155.
[3] 李雅.建筑平面布局的自动生成方法综述[J].安徽建筑,2021,28(8):158-161,182.
[4] Zhang H, Zhang R.Generative artificial intelligence (AI) in built environment design and planning:A state-of-the-art review[J].Progress in Engineering Science,2025,2(1):100040.
[5] Nauata N,Chang K H,Cheng C Y,et al.House-GAN:Relation al generative adversarial networks for graph-constrained hou se layout generation∥Proceedings of the European Conferen ce on Computer Vision(ECCV).Cham:Springer,2020.
[6] Li J,Xu T, Zhang J, et al. LayoutGAN: Generating graphic layouts with wireframe discriminator [C]∥Proceedings of the International Conference on Learning Representations (ICLR). New Orleans:OpenReview,2019.
[7] Yang L,Zhang Z,Song Y,et al.Diffusion models:A compreh ensive survey of methods and applications[J].ACM Computing Surveys,2023,56(4):Article79.
[8] Yang L, Zhang Z, Song Y, et al. Mastering text-to-image diffusion:Recaptioning,planning,and generating with multimo dal LLMs[EB/OL].(2024-01-22)[2025-05-05].

Copyright © 2025 张百慧

Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License