Machine learning knowledge‐driven neuroarchitecture for analysis of three‐dimensional boundary layer flow of Sisko nanofluid model with impact of magnetic field

Abstract This communication employs an advanced machine learning procedure to investigate the effects of magnetic fields and nanoparticles on the three‐dimensional boundary layer flow of a Sisko nanofluid (BLF‐SNF) model driven by a bidirectional stretching surface. The analysis incorporates Brownian motion and thermophoresis within the nanofluid model, considering the fluid to be electrically conductive under the influence of a constant magnetic field. By utilizing a deep learning approach based on the Nonlinear Autoregressive Exogenous (NARX) networks trained with Bayesian Regularization optimization, the mathematical framework of the BLF‐SNF model is analyzed, for magnetic Reynolds numbers, incorporating zero particle mass flux at the surface, Sisko nanofluid parameter, and Prandtl number based sundry scenarios. Learning curves with MSE around 10 −10 –10 −14 , error histograms with a mean value of the central bin around 10 −06 –10 −09 , and absolute error around 10 −04 to 10 −10 for the solution profiles for velocity, transverse component of velocity, temperature, and nanoparticle concentration confirm the correctness of the approach with reasonable accuracy. The results highlight the predictive accuracy of the machine learning model, revealing effects of Brownian motion and thermophoresis along with the impact of variation in the sundry physical quantities on trajectories of velocity, temperature, and nanoparticle concentration of the BLF‐SNF model.