A NEW STUDY ON THE PREDICTION OF THERMAL EFFICIENCY PROPERTIES OF OLDROYD-B NANOFLUID FLOW IN SOLAR WATER PUMPS WITH AN ARTIFICIAL NEURAL NETWORK MODEL WITH BAYESIAN REGULARIZATION ALGORITHM

Efficient use of the endless energy from the sun not only provides an effective solution to environmental problems, but also offers significant financial gains. Due to their strong thermal characteristics, the use of nanofluids in solar systems is becoming widespread. In this study, the effect of using two different nanofluids on thermal efficiency in solar water pumps was examined using an artificial neural network. Two separate nanofluid flows, based on engine oil and composed of copper and graphene oxide nanoparticles, were considered. A multilayer artificial neural network model was developed to predict the thermal efficiency parameters of both nanofluid flows. The Bayesian regularization training algorithm was used in neural network models with multilayer perceptron architecture. The output values obtained from the neural network were compared with the real values and a high agreement was observed. The coefficient of performance value for the neural network model was obtained as 0.95088 and the mean squared error value as 6.87E-05. This research, in which the thermal efficiency characteristics of engine oil-based nanofluid flow in a solar water pump system are examined with an artificial intelligence approach, shows the usability of artificial neural networks in predicting the parameters of nanofluid use in solar systems with high accuracy.