Artificial neural computing and statistical analysis of heat and mass transport of nanofluid flow with melting heat and thermal stratification

The melting heat phenomenon in viscous nanofluid MHD flow over a stretching sheet with variable porosity and permeability was utilized. The behavior of magneto-viscous nanofluid flow was scrutinized using the concepts of machine learning and statistics. Thermal stratification, heat source, and activation energy in the solution of tween-20 nanoparticles and an ethyl-acetate base-type fluid are examined. Brownian and thermophoretic phenomena are included. Non-linear partial differential equations (PDEs) are converted into non-linear ordinary differential equations (ODEs) via von Karman similarity variables. Dataset is generated by the 4th-order Runge-Kutta numerical method for the artificial neural network. The ratio parameter and melting parameter enhance the velocity outline, while the melting parameter and thermal stratification parameter reduce the temperature outline. An increase in the concentration outline is seen with the activation energy parameter and Brownian motion parameter both increase the concentration outline. The system's performance using metrics like regression analysis, mean squared error, and error histograms is evaluated. The impact of these factors on significant results, such as the drag coefficient and heat transfer rate, is statistically investigated using multiple linear regressions. The integration of statistical and machine learning methods is deemed crucial for enhancing understanding of complex fluid dynamics in magnetic nanofluid flows.