Forecasting Hot Water Consumption in Dwellings Using Artificial Neural Networks

Gelazanskas, L. and Gamage, K. A.A. (2015) Forecasting Hot Water Consumption in Dwellings Using Artificial Neural Networks. In: 2015 IEEE 5th International Conference on Power Engineering, Energy and Electrical Drives (POWERENG), Riga, Latvia, 11-13 May 2015, pp. 410-415. ISBN 9781479999781 (doi: 10.1109/PowerEng.2015.7266352)

Full text not currently available from Enlighten.


The electricity grid is currently transforming and becoming more and more decentralised. Green energy generation has many incentives throughout the world thus small renewable generation units become popular. Intermittent generation units pose threat to system stability so new balancing techniques like Demand Side Management must be researched. Residential hot water heaters are perfect candidates to be used for shifting electricity consumption in time. This paper investigates the ability on Artificial Neural Networks to predict individual hot water heater energy demand profile. Data from about a hundred dwellings are analysed using autocorrelation technique. The most appropriate lags were chosen and different Neural Network model topologies were tested and compared. The results are positive and show that water heaters have could potentially shift electric energy.

Item Type:Conference Proceedings
Glasgow Author(s) Enlighten ID:Gamage, Professor Kelum
Authors: Gelazanskas, L., and Gamage, K. A.A.
College/School:College of Science and Engineering > School of Engineering > Systems Power and Energy

University Staff: Request a correction | Enlighten Editors: Update this record