Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Civil-Comp Proceedings
ISSN 1759-3433
CCP: 74
Edited by: B.H.V. Topping and B. Kumar
Paper 28

Development of a Neural Network to Predict Residential Energy Consumption

R.R.A. Issa, I. Flood and M. Asmus

University of Florida, Gainesville, Florida, United States of America

Full Bibliographic Reference for this paper
R.R.A. Issa, I. Flood, M. Asmus, "Development of a Neural Network to Predict Residential Energy Consumption", in B.H.V. Topping, B. Kumar, (Editors), "Proceedings of the Sixth International Conference on the Application of Artificial Intelligence to Civil and Structural Engineering", Civil-Comp Press, Stirlingshire, UK, Paper 28, 2001. doi:10.4203/ccp.74.28
Keywords: neural network, energy consumption, energy performance index, building parameters, conditioned living area, environmental impact.

In Florida, an Energy Performance Index (EPI) calculation must be performed and submitted before a building permit can be granted. The EPI is a measure of energy efficiency calculated for all new construction and renovations. It uses data on several components of a structure to assign a rating (it must be under 100 points to pass). The lower the EPI, the more efficient the structure should be. This study involved the collection of residential EPI data from participating Gainesville Regional Utilities customers, matching this data with the actual energy consumption data and training an artificial neural network to relate the EPI to actual energy consumption. The ability of the artificial neural network to predict annual energy consumption will help residential designers/builders advance the goals of reducing the monthly cost of new housing and reducing the associated environmental impact and energy use. Data on new residential construction EPI calculations for 1998-2000 in Alachua County, Florida and corresponding energy consumption data for one year was compiled. This data was also matched with the conditioned living area of each house and then imported into the neural network. A local subdivision in which several houses were issued permits based on the same EPI was also identified as a control group so that normal variations in annual energy consumption could be determined. The neural network was used to create a model to predict annual energy consumption cost.

Previous predictive studies of EPI [1] on which this study is based have been successful in linking the most influential of the approximately 100 factors used to calculate EPI to the actual EPI score. The purpose of this research is to continue predictive studies of EPI by now linking it to outside factors. Particularly, this research attempts to create a model that links EPI to actual energy consumption. As with earlier research, this model was created using a back-propagation artificial neural network, as shown in Figure 28.1. Artificial neural networks have the ability to link given inputs and outputs by finding the correlation between them. Based on this, the neural network creates a pattern, or learns, from the data. Once it has learned how inputs and outputs are related, it can take a given input and return the correct output. In other words, neural networks have the ability to correctly classify or predict an outcome based on inputs it has not learned or "seen" before.

The objective of this research is to develop a neural network that allows the user, based on information about the conditioned living area of a given house and its EPI, to predict its annual energy consumption. This will create a predictive tool for future use in Central Florida (where the data was collected) with the possibility to replicate the same approach in other parts of the USA and the world.

Figure 28.1: Three-Layer Back-Propagation Neural Network Configuration

R.R.A. Issa, I. Flood, de Martini A., "Estimating the Energy Performance Index of Buildings" in I. Flood and N. Kartam (Eds.) Artificial Neural Networks For Civil Engineers, Reston, VA, ASCE , 260-272, 1998.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description