Prediction of viscous dissipation effects on magnetohydrodynamic heat transfer flow of copper-poly vinyl alcohol Jeffrey nanofluid through a stretchable surface using artificial neural network with Bayesian Regularization

In this study, the viscous dissipation effects of copper-polyvinyl alcohol (Cu-PVA) Jeffrey nanofluid on magnetohydrodynamic (MHD) heat transfer flow across a stretchable surface have been analyzed with an artificial intelligence approach. The flow parameters, skin friction and Nusselt number, are numerically obtained with a closed Keller-box and partial differential equations converted to a non-linear ordinary differential equation system using the appropriate similarity transformation. Using the obtained data set, two different artificial neural network (ANN) models have been developed. In the multi-layer perceptron (MLP) network model developed with Bayesian Regularization training algorithm, solid volume fraction (φ), Deborah number (β), magnetic parameter (M), Prandtl number (Pr) and Eckert number (Ec) values have been defined as input parameters and skin friction and Nusselt number values ​​have been obtained in the output layer. R values ​​for skin friction and Nusselt number have been calculated as 0.99020 and 0.99394, respectively. The study findings show that the developed ANN model can predict with high accuracy and is a high-performance engineering tool that can be used in modeling viscous dissipation effects.