Proceedings of the 2022 European Conference on Computing in Construction


PRENERGET: A framework for the inclusion and adaptation strategies of machine learning models for energy demand forecasting in buildings

Iker Esnaola-Gonzalez, Meritxell Gomez-Omella, Susana Ferreiro, Francisco Javier Diez, Alvaro García
Tekniker, Spain

DOI: 10.35490/EC3.2022.149
Abstract: Machine Learning (ML) models are key enablers for the implementation of different energy-efficiency strategies in buildings. There are a variety of frameworks that facilitate the development of ML models, but it is necessary to move into a different environment for their deployment and exploitation. Furthermore, their performance tends to degrade over time. Consequently, they need to be regularly evaluated and upgraded to ensure the robustness of the overall solution. The seamless exploitation, adaptation and evolution of ML models is still an open issue nowadays, and in this article, a software framework called PRENERGET aimed at addressing this issue is presented.
Keywords: energy demand, forecasting, machine learning, buildings

Presentation Video: