Central University of Technology, P/SAK X20539, Bloemfontein, 9301, South Africa

---

Abstract: Air to water heat pump (AWHP) is an efficient and renewable technology for sanitary water heating. The study focused on the building and development of an artificial neural network (ANN) to predict the electrical energy consumed (E) and COP of a 1.2 kW split type AWHP with the volume of hot water drawn, ambient temperature, relative humidity, difference in refrigerant temperatures at the outlet and inlet of the compressor and at the inlet and outlet of the condenser as the input parameters. An ANN of 5-10-2 configuration with Levenberg-Marquardt as the variant of the back propagation algorithm was used to train the input and output dataset. The trained network shows that both the modelled outputs and the targets of the AWHP for the summer season mimic each other with a deviation of ±0.019. The correlation coefficients (R) for the training, validation and testing sample dataset with the trained network was 0.967, 0.962 and 0.945, respectively. The trained ANN was used to evaluate the network with an additional test dataset of the inputs and outputs that were not considered during the training of the ANN. The modelled outputs and targets for the evaluation network gave an excellent prediction with a correlation coefficient and mean square error of 0.996 and 0.003, respectively. We can conclude that the trained ANN is simple to configure and less time consuming in building and training the network, but, was capable of predicting both the E and COP of the AWHP with reasonably high accuracy with a 95% confidence level.

---

Full Paper.pdf

Copyright © 2022 by the authors. This is an open access article distributed under the Creative Commons Attribution License (CC BY-NC-ND 4.0), which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.