Designing an artificial neural network to predict dynamic viscosity of aqueous nanofluid of TiO2 using experimental data

In this research, the viscosity of the aqueous nanofluid of TiO2 has been modeled by artificial neural networks using experimental data. Artificial neural networks are able to estimate the pattern of dynamic viscosity variation along with temperature and nanoparticles mass fraction with a high precision. A network with one hidden layer and 4 neurons has been used. The regression coefficient was obtained 0.9998 in this modeling, which shows very high precision of neural network with a very simple structure. In addition, a relationship in terms of mass fraction and temperature was presented in order to predict the viscosity of this nanofluid. This correlation can estimate the viscosity of TiO2-water nanofluid in a wide range of nanoparticles mass fraction with a maximum error of 0.5 %.