Application of Artificial Neural Networks onBuilding Energy Estimation

Abstract -The decision makers in the building design stage need to give more importance for energy efficient solutions to improve the performance the building in its operational phase. In practice, they are expected to generate alternative designs in the early design stage and use complicated building simulation tools to estimate building energy use of these alternatives more accurately; however, this is a time-consuming procedure and it is difficult to evaluate different alternatives in the limited time of a design period. For this reason, emerging artificial intelligence (AI) algorithms are becoming popular amongst practitioners in the industry to study the energy performance evaluation. Their inherently fast nature makes AI algorithms suitable for this topic depending on the database generated. In this study, an Artificial Neural Network (ANN) is developed to estimate energy performance of the buildings efficiently based on a previously generated dataset including building properties and energy loads. The ANN is constructed based on different network topologies, activation functions and learning rates combinations to provide more accurate results on heating and cooling loads. The results of the study are presented and compared with the results in the previous studies. It shows the right parameter combination for ANN generates quite competitive results with the ones from the previous studies.
Citation Formats
M. Altun, A. B. Ersöz, T. Teke, T. Kurt, A. Akçamete Güngör, and O. Pekcan, “Application of Artificial Neural Networks onBuilding Energy Estimation,” presented at the International Conference on Engineering Technologies, (7 - 09 Aralık 2017), Konya, Türkiye, 2017, Accessed: 00, 2021. [Online]. Available: