Partager

Quantification basée sur l’apprentissage profond des gains de chaleur et impact sur la demande énergétique des bâtiments.

Deep learning-based quantification of heat gains and impact on building energy demand.

Numéro : 3575

Auteurs : MAH D., CHAI H., KIRCHER K. J., TZEMPELIKOS A.

Résumé

This paper introduces a novel approach for real-time monitoring of dynamic internal and solar heat gains using programmable low-cost cameras and deep learning techniques. A convolutional neural network (CNN)-based multi-head classification model was trained with High Dynamic Range (HDR) images, collected using a low-cost fisheye camera in a private office and fine-tuned using a separate dataset from an open-plan office. The results showed that the developed model could classify the status of multiple heat gains (occupants, equipment, lighting, windows) in predefined areas of the scene with great performance, achieving high precision and recall results. Furthermore, to evaluate the impact of real-time heat gain monitoring on energy demand, the large office space was modeled with energy simulation software using commonly assumed fixed heat gain schedules and real-time monitored dynamic schedules under the same weather conditions. The results showed that using fixed schedules may lead to significant errors, resulting in underestimation of some thermal load components and overestimation of others.

Documents disponibles

Format PDF

Pages : 8 p.

Disponible

Gratuit

Télécharger

Détails

  • Titre original : Deep learning-based quantification of heat gains and impact on building energy demand.
  • Identifiant de la fiche : 30032908
  • Langues : Anglais
  • Source : 2024 Purdue Conferences. 8th International High Performance Buildings Conference at Purdue.
  • Date d'édition : 15/07/2024

Liens


Voir d'autres communications du même compte rendu (63)
Voir le compte rendu de la conférence

Indexation