AUTHOR=Lin Mengyuan , Peng Liyuan , Liu Tingting , Zhang Lili TITLE=Novel approach to energy consumption estimation in smart homes: application of data mining and optimization techniques JOURNAL=Frontiers in Energy Research VOLUME=12 YEAR=2024 URL=https://www.frontiersin.org/journals/energy-research/articles/10.3389/fenrg.2024.1361803 DOI=10.3389/fenrg.2024.1361803 ISSN=2296-598X ABSTRACT=

Buildings account for a significant portion of total energy consumption, and the introduction of intelligent buildings represents a significant step forward in efficiently managing energy utilization. The proposed solutions represent a significant step forward in the development of intelligent residential environments. Beginning the process of achieving improved building intelligence necessitates a thorough evaluation and prediction of the necessary heating and cooling energy requirements, taking into account all relevant influencing factors. This study describes methodologies for using data mining models to predict the heating and cooling energy requirements of intelligent buildings during the construction phase. Data mining techniques, specifically Support Vector Machines (SVM) and Random Forest, are used, demonstrating their superior efficiency over alternative methods. Metaheuristic algorithms, particularly the Owl Search Algorithm (OSA), are described as effective tools for optimizing results across a wide range of problem resolutions. OSA is described and proposed alongside novel data mining methods, demonstrating that this combination of algorithms improves the performance of Random Forest and SVM-based models by 11% and 24%, respectively. The proposed models can generate predictions with a small number of parameters, eliminating the need for complex software and tools. This user-friendly approach makes the prediction process more accessible to a wider audience. While specialized equipment and professional-grade tools will be used, the proposed models are accessible to a wide range of individuals interested in participating in the prediction process.