Artificial neural networking for computational assessment of ternary hybrid nanofluid flow caused by a stretching sheet: implications of machine-learning approach

Researchers are mainly interested in using soft computing artificial intelligence (AI) methods due to their broad applications in analysis, modelling and simulations. Backpropagation neural networks, one of the supervised learning algorithms, is commonly used to train data networks by optimizing the error between actual and predicted values. To optimize this process of training data, Levenberg-Marquardt algorithm is applied; particularly beneficial for solving nonlinear fluid flow problems. Little knowledge is known about the ternary-hybrid nanofluid flow caused by a stretching surface with heat generation, viscous dissipation, magnetic effect and porosity etc. This article presents a novel machine-learning approach using backpropagation neural networks augmented with the Levenberg-Marquardt procedure to figure out the ternary-hybrid nanofluid flow generated by a stretching sheet. It uniquely examines the mixture of copper, Iron oxide and silicon dioxide nanoparticles inside a single base fluid within a magnetic field, tackling the research gaps in the effects of heat generation, viscous dissipation, porosity, and magnetic effects on fluid flow system and heat transfer. Shooting numerical technique (RK-5th) is used for solving the governing ordinary equations. Graphical illustration, error analysis, mean squared error, histograms and regression analysis justify the proposed method, showing better performance for ternary nanofluids.