Speaker
Description
Modelling is a power tool in a development of an engineering system. In this study, a solar vapour compression refrigeration system was modelled by two machine learning approaches, namely ARX and ANN and the performance of these two approaches were compared. The solar refrigeration system is composed mainly of a vapour compression unit, two 300W-solar modules, two 12V-batteries with the capacity of 200 Ah (each) and a charge controller. Ten experiments were carried out. Water contained in a bottle was used as cooling loads. The load of 10, 20, 30, 40, 50, 60, 70, 80, 90 and 100 liters were employed in the experiments. Data obtained from the experiments with the loads of 10, 30, 50, 70 and 100 liters of water were used to construct the model and experiments with the loads of 20, 40, 60, 80 and 90 liters of water were employed to test the performance of the model. The difference of the load temperature from the experiments and those predicted by the models, in terms of the percentage of root mean square difference relative to a mean measured values (RMSD) was used as an indicator of the performance of the models. It was found that the RMSD of ARX model and ANN model were 4.3% and 5.4%, respectively. We conclude that ARX approach performs batter than ANN approach for this system.
