| id |
caadria2025_92 |
| authors |
Chen, Kevin, Liu, Yubo and Deng, Qiaoming |
| year |
2025 |
| title |
Optimizing the Novel Educational Buildings for Daylighting and Outdoor Platform through Deep Reinforcement Learning |
| source |
Dagmar Reinhardt, Christiane M. Herr, Anastasia Globa, Jielin Chen, Taro ?Narahara, Nicolas Rogeau (eds.), ARCHITECTURAL INFORMATICS - Proceedings of the 30th CAADRIA Conference, Tokyo, 22-29 March 2025, Volume 3, pp. 183–192 |
| summary |
Amid the COVID-19 pandemic, public awareness of the relationship between the built environment and health has grown significantly. Classrooms, as primary learning spaces, play a crucial role in students' learning and physical health. Inadequate daylighting causes visual discomfort, impairing vision and learning efficiency, while limited outdoor space heightens mental health risks by reducing opportunities for intersessional activities. This study proposes a composite open-air classroom that maximizes daylighting and outdoor activity spaces, drawing inspiration from the historical "Open-Air School". Multi-Agent Deep Reinforcement Learning (MADRL) is introduced to address the complex decision-making process across multiple classrooms. Six metrics—Spatial Daylight Autonomy (sDA), Daylighting of Uniformity (UOD), Annual Sunlight Exposure (ASE), Average Predicted Mean Vote (APMV), Outdoor Platform Area (OPA), and Space Utilization (SU)—are utilized to evaluate building performance and spatial layout. To enhance training efficiency, Artificial Neural Networks (ANN) are employed to develop predictive models for time-consuming indicators. As a result, MADRL shows better performance and produces more diverse outcomes. This study advances both the theory and practice of educational building design, promoting a more human-centered and health-conscious future. |
| keywords |
Reinforcement Learning, Daylighting, ANN, Outdoor Platform, Educational Building |
| series |
CAADRIA |
| full text |
 file.pdf (2,805,294 bytes) |
| references |
Content-type: text/plain
|
| last changed |
2025/03/21 12:08 |
|
