Artificial neural networks (ANNs) are a commonly used and effective method for estimating the outputs of nonlinear plants. In this paper, a neural network is used to estimate the electricity consumption of Cooper Union’s new high performance academic building at 41 Cooper Square. Although neural networks have been used for over 20 years to estimate municipal or regional energy consumption, extending their application to analyzing the energy consumption of individual buildings offers great potential. Because most existing ANN models are implemented over a municipal or regional scale, additional variables were considered as inputs to this model to account for factors specific to individual buildings. Depending on a building’s height and aspect ratio, the location of the sun can have a significant effect on heating and cooling loads. This paper analyzes the effect of cloud cover and solar position on the accuracy of ANN models and finds that the presence of solar position as an input can decrease the predicted energy consumption error by as much as 48%.
Design of Artificial Neural Network Using Solar Inputs for Assessing Energy Consumption in a High Performance Academic Building
- Views Icon Views
- Share Icon Share
- Search Site
Sterman, M, & Baglione, M. "Design of Artificial Neural Network Using Solar Inputs for Assessing Energy Consumption in a High Performance Academic Building." Proceedings of the ASME 2012 International Mechanical Engineering Congress and Exposition. Volume 10: Emerging Technologies and Topics; Public Policy. Houston, Texas, USA. November 9–15, 2012. pp. 143-149. ASME. https://doi.org/10.1115/IMECE2012-86976
Download citation file: