Neuro-computing intelligent networks to analyze Casson nanofluid flow over a curved stretching surface

Abstract

In order to provide acceptable exact estimates for Casson nano-fluid flow over stretching surface (CNF-CSS), this study designed an effective approach Levenberg-Marquardt technique artificial neural networks (LMTs-ANNs). LMT back-propagation is quick for nonlinear least-squares problems. A quasi-network of PDEs expressing CNF-CSS needs adjustments to become standards. We create datasets to train, test, and verify the LMTs-ANNs model against the solution of the bvp4c technique. The practical data is derived using the mean square error (MSE) and absolute error (A.E) assessment ratios. For achieving consistency, the IBPNNs-LMTs architecture is employed to execute intellectual development. The usefulness of the produced IBPNNs-LMTs for such a modelled issue is demonstrated by achieving the most promising mathematical outcomes in the range of E-03 to E-08, as well as measurements of error-histogram analysis (E.H.A) and regression analysis (R.A). Mu is a controller that oversees the entire training procedure. This study also discusses several dimensionless parameters, such as Sc, the momentum-mass diffusivity relationship. A rise in Sc lowers particle diffusivity, decreasing the concentration gradient in terms of the physical phenomena. Getting the most attractive mathematical results for A.E in the range of demonstrates the applicability of the generated IBPNNs-LMTs for such a modelled situation.

KEYWORDS:

• CNF-CSS
• LMTs-ANNs
• artificial neural networks
• BVP4C

Acknowledgments

This study is supported via funding from Prince Sattam bin Abdulaziz University project number (PSAU/2023/R/1444).

Disclosure statement

No potential conflict of interest was reported by the author(s).