Utilizing artificial neural network to predict energy consumption and thermal comfort level: An indoor swimming pool case study

This paper presents an ANN approach to predict energy consumption and thermal comfort level of an indoor swimming pool. In Swimming pool, several environmental and control variables, directly or indirectly, affect energy consumption and thermal comfort, rendering difficult the development of a mathematical relationship amongst input and output variables. Thus, ANN based prediction approach is used to elicit this relationship within reasonable period of time. This forms the basis of an optimization based control system to evaluate the control parameters in the swimming pool. The proposed approach is implemented for a specific HVAC system, based on scenarios developed in close consultation with site engineers and domain experts. Due to lack of meaningful historical monitored data, a calibrated simulation model is used to generate large amount of data sets to train the corresponding ANN prediction engine. The trained ANN was then calibrated in real conditions and used as a cost function in an optimization program to help achieve energy saving targets. Several ANN algorithms have been tested and benchmarked leading to the selection, with further tuning, of the best performing ANN algorithm, namely Levenberg-Marquardt algorithm. The latter was used and achieved good results as demonstrated in the selected case study. (C) 2014 Elsevier B.V. All rights reserved.