Design of Ensemble Forecasting Models for Home Energy Management Systems

• Main Author: Bot, Karol
• Publication Date: 2021
• Other Authors: Santos, Samira, Laouali, Inoussa, Ruano, Antonio, Ruano, Maria da Graça
• Format: Article
• Language: eng
• Source: Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
• Download full: https://hdl.handle.net/10316/103878; https://doi.org/10.3390/en14227664
• topic: energy systems; machine learning; forecasting; energy management systems; multiobjective genetic algorithms; ensemble models; energy in buildings

The increasing levels of energy consumption worldwide is raising issues with respect to surpassing supply limits, causing severe effects on the environment, and the exhaustion of energy resources. Buildings are one of the most relevant sectors in terms of energy consumption; as such, efficient Home or Building Management Systems are an important topic of research. This study discusses the use of ensemble techniques in order to improve the performance of artificial neural networks models used for energy forecasting in residential houses. The case study is a residential house, located in Portugal, that is equipped with PV generation and battery storage and controlled by a Home Energy Management System (HEMS). It has been shown that the ensemble forecasting results are superior to single selected models, which were already excellent. A simple procedure was proposed for selecting the models to be used in the ensemble, together with a heuristic to determine the number of models.