Marangoni convection-driven squeezing flow and thermal transport of water-based pseudoplastic ternary hybrid nanofluid in a porous channel: machine-learning optimization

In this study, an unsteady Marangoni convection squeezing flow of a pseudoplastic ternary-HNF driven by a temperature gradient is investigated by means of numerical and neural network predictions. The influences of Marangoni convection, radiative heat flux and cross diffusion are scrutinized. The governing problem is solved numerically using collections methods in order to obtain reference data for ANN training. To achieve the best degree of machine learning, an artificial neural network (ANN) based on the Levenberg–Marquardt is trained on numerical data. These findings indicate that the flow of ternary-HNF declines with Forchheimer, Darcy and power law numbers and the volume fraction of nanomaterials. The thermal transport uplifts with Forchheimer, Darcy, Dufour, power law number and the volume fraction of nanomaterials. The mass of a species declines with Forchheimer and Soret number. The proposed ANN hybrid model exhibits a high degree of alignment with extremely low mean squared errors (MSE) 1.2387×10−10 to 1.3178×10−9.