Neural Network Analysis of Soret Effect on Rotating Fluid Between Parallel Plates With Magnetohydrodynamic Over Porous Media

ABSTRACT This study investigates the influence of the Soret effect on unsteady, two‐dimensional magnetohydrodynamic (MHD) heat and mass transfer in a rotating fluid confined between parallel porous plates under a transverse magnetic field and thermal radiation. The primary objective is to quantitatively predict flow behavior using a hybrid approach combining numerical simulation with artificial neural networks (ANNs). Governing partial differential equations are non‐dimensionalized via similarity transformations and solved numerically using MATLAB's bvp4c solver. A deep‐learning model is then trained on the synthetic data set to validate and forecast velocity, temperature, and concentration profiles. Key findings reveal that an increase in the Prandtl number suppresses thermal diffusion, while the Soret number enhances mass transfer. The ANN predictions exhibit high accuracy ( R ² > 0.94), confirming the robustness of the data‐driven methodology. This study demonstrates the efficacy of machine learning in capturing nonlinear transport phenomena in rotating porous MHD systems, offering a computationally efficient alternative for industrial thermal‐fluid design.