https://doi.org/10.1140/epjp/s13360-024-05687-x
Regular Article
Modeling the indoor temperature depending on insulation thickness using machine learning methods
1
Engineering Faculty Computer Engineering Department, Sivas Cumhuriyet University, 58140, Sivas, Türkiye
2
Sivas Vocational High School, Sivas Cumhuriyet University, 58140, Sivas, Türkiye
3
Engineering Faculty Mechanical Engineering Department, Sivas Cumhuriyet University, 58140, Sivas, Türkiye
4
Sivas Cumhuriyet University, 58140, Sivas, Türkiye
Received:
2
May
2024
Accepted:
25
September
2024
Published online:
9
October
2024
The efficient use of energy is of paramount importance for the sustainability of the global energy sector. A building’s energy requirements are contingent upon a multitude of factors, including the surrounding environment and meteorological conditions, the specific construction materials utilized, the level of thermal insulation incorporated, and the comfort preferences of the building’s occupants. It is of paramount importance to consider all of these factors when designing and constructing buildings in order to ensure that they are as energy-efficient as possible. The primary objective of this study is to develop machine learning models that can accurately predict the indoor temperature of a thermally insulated building, with the aim of improving energy efficiency. As a novel approach, a variety of machine learning techniques, including artificial neural networks (ANN), Gaussian process regression (GPR), support vector machines (SVM), and adaptive network-based fuzzy inference systems (ANFIS), are employed and a comparative analysis is conducted based on insulation thickness, heating settings, and outdoor temperature values. The data set was obtained from experimental measurements collected during the heating season in an insulation house constructed on a university campus in Sivas, Turkey. The data set comprises parameters such as insulation thickness, heating settings, and outdoor temperature values, which were collected during the model development process. The results of the study demonstrate that the test statistical metric values of R2, RMSE, MAPE, and MAE are 0.862, 0.263, 0.009, and 0.190, respectively, for the ANFIS model and 0.997, 0.149, and 0.005, respectively, for the ANN model. The ANN model yielded the following results: 0.107. The SVM model yielded the following results: 0.936, 0.601, 0.020, and 0.449. The GPR model yielded the following results: 0.998, 0.134, 0.004, and 0.074. The results demonstrate that the GPR and ANN models exhibit superior accuracy and lower error metrics compared to the other models. Notably, the GPR model was identified as the most successful model with the lowest RMSE and MAPE values, indicating that the model performs optimally and is an effective approach that can be applied in this field. This study aims to contribute to sustainable energy use by emphasizing the effectiveness of machine learning techniques in energy management of buildings.
Copyright comment Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
© The Author(s), under exclusive licence to Società Italiana di Fisica and Springer-Verlag GmbH Germany, part of Springer Nature 2024. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
