Dynamic System Testing Method and Artificial Neural Networks For Solar Water Heater Long-Term Performance Prediction

Abstract

The performance of a solar hot water thermosyphon system was tested with the dynamic system method according to Standard ISO/CD/9459.5. The system is of closed circuit type and consists of two flat plate collectors with total aperture area of 2.74 m2 and of a 170 liters hot water storage tank. The system was modeled according to the procedures outlined in the standard with the weather conditions encountered in Rome. The simulations were performed for hot water demand temperatures of 45 and 90°C and volume of daily hot water consumption varying from 127 to 200 liters. These results have been used to train a suitable neural network to perform long-term system performance predictions. A total of 5 complete runs (60 patterns) were available. From these, 12 patterns with data for a whole year were used as a validation set, whereas the rest 48 were used for the training (42 sets) and testing (6 sets) of the network. A multi layer feedforward neural network with three hidden slabs was used. Seven input and four output parameters are used. The input data were leaned with adequate accuracy with correlation coefficients varying from 0.993 to 0.998, for the four output parameters. When unknown data were used to the network, satisfactory results were obtained. The maximum percentage difference between the actual (simulated) and predicted results is 6.3%. These results prove that artificial neural networks can be used successfully for this type of predictions.