International conference on construction engineering and project management (국제학술발표논문집)

  • 2024.07a
  • /
  • Pages.973-979
  • /
  • 2024
  • /
  • 2508-9048(eISSN)
Korea Institute of Construction Engineering and Management (한국건설관리학회)
권호 표지
DOI QR코드

DOI QR Code

Mixed-input neural networks for daylight prediction

  • Thanh Luan LE (Department of Global Smart City, Sungkyunkwan University) ;
  • Sung-Ah KIM (Department of Architecture, Sungkyunkwan University)
  • Published : 2024.07.29
Download PDF
⟨ Previous Next ⟩

Abstract

In this research, we present the implementation of a mixed-input neural network for daylight prediction in the architectural design process. This approach harnesses the advantages of both image and numerical inputs to construct a robust neural network model. The hybrid model consists of two branches, each handling in-depth information about the building. Consequently, this model can effectively accommodate a wide range of building layouts, incorporating additional information for enhanced predictions. The building data was created utilizing PlanFinder in Rhino Grasshopper, while simulation data were generated using Honeybee and Ladybug. Weather data were collected from three distinct localities in Vietnam: Ha Noi, Da Nang, and Ho Chi Minh City. The neural network demonstrates outstanding performance, achieving an R-squared (R2) value of 0.95 and the overall percentage difference in the testing dataset ranges from 0 to 20.7%.

Keywords

Acknowledgement

This work is financially supported by Korea Ministry of Land, Infrastructure and Transport (MOLIT) as Innovative Talent Education Program for Smart City and supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. 2022R1A2C1010759).

References

  1. M. Ayoub, "100 Years of daylighting: A chronological review of daylight prediction and calculation methods," Solar Energy, vol. 194, no. October, pp. 360-390, 2019, doi: 10.1016/j.solener.2019.10.072.
  2. Q. Liu, Y. Chen, Y. Liu, Y. Lei, Y. Wang, and P. Hu, "A review and guide on selecting and optimizing machine learning algorithms for daylight prediction," Build Environ, p. 110822, Oct. 2023, doi: 10.1016/j.buildenv.2023.110822.
  3. T.-L. Le, H. Nguyen-Xuan, and S.-A. Kim, "Hand drawing-based daylight analysis using deep learning and augmented reality," Results in Engineering, vol. 22, p. 102119, Jun. 2024, doi: 10.1016/j.rineng.2024.102119.
  4. T.-L. Le and S.-A. Kim, "Game-based Platform for Daylight Analysis using Deep Learning," in Proceedings of the International Conference on Education and Research in Computer Aided Architectural Design in Europe, 2023, pp. 481-490. doi: 10.52842/conf.ecaade.2023.2.481.
  5. L. Le-Thanh, H. Nguyen-Thi-Viet, J. Lee, and H. Nguyen-Xuan, "Machine learning-based real-time daylight analysis in buildings," Journal of Building Engineering, vol. 52, no. November 2021, p. 104374, Jul. 2022, doi: 10.1016/j.jobe.2022.104374.
  6. M. Ayoub, "A review on machine learning algorithms to predict daylighting inside buildings," Solar Energy, vol. 202, no. January, pp. 249-275, May 2020, doi: 10.1016/j.solener.2020.03.104.
  7. C. L. Lorenz, M. Packianather, A. Benjamin Spaeth, and C. B. De Souza, "Artificial Neural NetworkBased Modelling for Daylight Evaluations," in Proceedings of the 2018 Symposium on Simulation for Architecture and Urban Design (SimAUD 2018), Society for Modeling and Simulation International (SCS), 2018, pp. 8-15. doi: 10.22360/SimAUD.2018.SimAUD.002.
  8. C. H. Lin and Y. S. Tsay, "A metamodel based on intermediary features for daylight performance prediction of facade design," Build Environ, vol. 206, Dec. 2021, doi: 10.1016/j.buildenv.2021.108371.
  9. Q. He et al., "Predictive models for daylight performance of general floorplans based on CNN and GAN: A proof-of-concept study," Build Environ, vol. 206, p. 108346, Dec. 2021, doi: 10.1016/j.buildenv.2021.108346.
  10. "PlanFinder." Accessed: Jan. 20, 2024. [Online]. Available: https://www.planfinder.xyz/
  11. A. N. American, "ANSI/IES LMM83322 APPROVED METHOD: IES SPATIAL DAYLIGHT AUTONOMY SDAA AND ANNUAL SUNLIGHT EXPOSURE ASEE." [Online]. Available: www.ies.org
  12. J. M. Twomey and A. E. Smith, "Performance measures, consistency, and power for artificial neural network models," Math Comput Model, vol. 21, no. 1-2, pp. 243-258, 1995, doi: 10.1016/0895-7177(94)00207-5.