Artificial Neural Network for Predicting Heating Load in Energy-Efficient Buildings
ADBA Computer Science · 30.07.2026 · 3 (2) · 81-87
DOI: 10.69882/adba.cs.2026072 ↗Kurz gefasst
Fachartikel in der Zeitschrift ADBA Computer Science, online erschienen am 30.07.2026. Verfasst von Mehmet Akif Koç. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).
Abstract (Originalsprache)
Precise forecasting of building energy requirements is crucial for enhancing energy efficiency and facilitating sustainable architectural design. This paper develops a machine learning strategy to forecast the heating load (kWh/m²) of residential buildings using the well-known Energy Efficiency dataset, comprising 768 samples and eight input variables describing building geometry and design parameters. The dataset was randomly partitioned into training (80%) and testing (20%) subsets, including 614 and 156 samples, respectively. A feedforward Artificial Neural Network (ANN) with a single hidden layer was used to model the nonlinear relationship between input variables and heating demand. A parametric study was conducted to determine the optimal model structure by varying the number of hidden-layer neurons (HLNN = 5, 15, 25, and 35). The model's performance was assessed by the coefficient of determination (R²), mean squared error (MSE), and mean error percentage (MEP). The findings demonstrate that the ANN model attained elevated prediction accuracy across all configurations. Optimal performance was achieved with HLNN set to 35, yielding R² values of 0.99795 for the training dataset and 0.99609 for the test dataset. The associated MSE values were 0.21457 and 0.36658, but the MEP values were −0.04618 and 0.32088. The close correlation between training and test outcomes indicates robust generalization without significant overfitting. These results validate that a basic ANN design may proficiently represent intricate nonlinear interactions in building energy systems. The proposed machine learning framework offers a precise, computationally efficient solution for predicting heating loads, aiding the design and optimization of energy-efficient buildings.
Zitieren
APA
Koç, M. A. (2026). Artificial Neural Network for Predicting Heating Load in Energy-Efficient Buildings. ADBA Computer Science, 3(2), 81-87. https://doi.org/10.69882/adba.cs.2026072
Vancouver
Koç MA. Artificial Neural Network for Predicting Heating Load in Energy-Efficient Buildings. ADBA Computer Science. 2026;3(2):81-87. doi:10.69882/adba.cs.2026072
BibTeX
@article{Ko2026,
author = {Koç, Mehmet Akif},
title = {Artificial Neural Network for Predicting Heating Load in Energy-Efficient Buildings},
journal = {ADBA Computer Science},
year = {2026},
volume = {3},
number = {2},
pages = {81--87},
doi = {10.69882/adba.cs.2026072}
}
EUVIMED