Forecasting of heat and mass transfer in Casson nanofluid flow with entropy optimization: Machine learning approach

Abstract The goal of this research is to deploy artificial neural networks (ANNs) to exploit the bioconvective flow of Casson nanofluid across a heated Riga plate with swimming microbes. The ANNs furnish a versatile computational framework that is very helpful in intricate domains, including biotechnology and biological computing. The impressions of nonlinear thermal radiation, heat generation/consumption and viscous dissipation are also addressed. In this study, thermophoresis and Brownian motion are modeled using the well-known Buongiorno nanofluid framework. The suitable alteration is provided to convert the governing flow expressions into an ordinary differential equations. Analytical solutions are derived with the homotopy analysis method. Our outcomes are consistent with the results reported in previous literature. The physical parameters, accompanied by modeled equations, are illustrated graphically alongside corresponding physical outcomes. The graphical values of velocity, thermal, nanofluid concentration, microorganisms, surface drag force, heat transmittal gradient, mass transmittal gradient, and motile density microorganisms are calculated and detailed for various emerging factors. The velocity field improved when mounting the quantities of mixed convection parameter and modified Hartmann number. The microorganisms field declines as the quantity of bioconvective Lewis number upsurges. The heat transmittal rate slumps when enhancing the value of the porosity parameter and the reverse trend is obtained for the mass transmittal rate. The Levenberg–Marquardt approach is employed to optimize the data by training (60%), testing (20%) and validating (20%) to verify the data’s accuracy.