Apprentissage automatique informé pour développer un modèle d'ordre réduit d'équipement unitaire.

Informed machine learning to develop a reduced order model of unitary equipment.

Numéro : 2386

Auteurs : YOUSAF S., BRADSHAW C. R., KAMALAPURKAR R., SAN O.

Résumé

Combining machine learning tools with conventional reduced order modeling approaches produces the potential for an enormous increase in the ability to select suitable models. This paper presents a machine learning based approach for predicting the cooling capacity of a fixed speed unitary air-conditioner. Experimental data from a 10-ton Roof Top Unit (RTU) is simplified by utilizing a feature selection methodology, Elastic Net (EN), to accurately record and reduce the parameters while simultaneously preserving the physics of the system. The simplified RTU data set resulting from the EN is fed into a novel method for model order reduction employing Principal Component Analysis coupled with a supervised Artificial Neural Network (ANN). Preliminary results show that proposed technique is able to predict the equipment cooling capacity within ±2% of experimental results. Furthermore, the current work has also been compared to the recent models in the literature and has been found to be superior.

Documents disponibles

Format PDF

Pages : 11 p.

Disponible

Gratuit

Détails

  • Titre original : Informed machine learning to develop a reduced order model of unitary equipment.
  • Identifiant de la fiche : 30030706
  • Langues : Anglais
  • Sujet : Technologie
  • Source : 2022 Purdue Conferences. 19th International Refrigeration and Air-Conditioning Conference at Purdue.
  • Date d'édition : 2022

Liens


Voir d'autres communications du même compte rendu (224)
Voir le compte rendu de la conférence