| id |
acadia24_v2_40 |
| authors |
Wang, Yixi; Hu, Xianglei |
| year |
2024 |
| title |
Optimization Research on the Hygrothermal Performance of Bamboo Woven Mud Walls Based on the GA-BP Neural Network with Small Sample Data |
| source |
ACADIA 2024: Designing Change [Volume 2: Proceedings of the 44th Annual Conference for the Association for Computer Aided Design in Architecture (ACADIA) ISBN 979-8-9891764-8-9]. Calgary. 11-16 November 2024. edited by Alicia Nahmad-Vazquez, Jason Johnson, Joshua Taron, Jinmo Rhee, Daniel Hapton. pp. 51-59. |
| summary |
Bamboo woven mud walls, an ancient construction type, hold rich historical and cultural significance. Traditionally used in various regions, these walls are known for their unique combination of bamboo and mud, which provide both structural support and thermal insu-lation. This study examines the hygrothermal performance of the traditional materials of bamboo woven mud walls, focusing on how well these traditional materials manage mois-ture and heat. To enhance the research data, a Generative Adversarial Network (GAN) is employed for data augmentation, effectively generating a more comprehensive dataset. The augmented data is then used to train a Back Propagation (BP) neural network to model the hygrothermal performance of the walls accurately. The application of Genetic Algorithm (GA) for model optimization reveals that the GA-optimized BP neural network model slightly outperforms the basic BP model in prediction accuracy.The results of this approach indicate that the GA-optimized BP neural network model exhibits slightly better prediction accu¬racy compared to the standard BP model. This suggests that integrating advanced machine learning techniques can improve the modeling and understanding of traditional construction materials, paving the way for their enhanced application in sustainable and energy-efficient building practices. |
| series |
ACADIA |
| type |
paper |
| email |
|
| full text |
 file.pdf (2,981,155 bytes) |
| references |
Content-type: text/plain
|
| last changed |
2025/07/21 11:41 |
|
