Design of intelligent computing networks for numerical treatment of thin film flow of Maxwell nanofluid over a stretched and rotating surface

Abstract

In this study, a novel application of backpropagated Levenberg-Marquardt neural networks (LM-NN) based computational intelligent heuristics is presented to interpret the analysis of Maxwell nanofluid thin film flow over a stretchable and rotating disk by considering magnetic and non-linear thermal radiation effects. Utilizing the Buongiorno model, thermophoretic and Brownian motion features of nanofluid are captured. The mathematical model in terms of partial differential equations (PDEs) is reduced to ordinary differential equations (ODEs) by incorporating the similarity transformations. The Adams numerical technique is utilized for generation of a dataset for proposed LM-NN in case of sundry scenarios of TFFPMN by variation of Deborah number, thermophoresis number, Schmidt number, radiation and Brownian motion variables. The training, testing and validation of the intelligent solver LM-NN is performed to find the solution of TFFPMN for various scenarios. Comparison with standard solution verified the precision of LM-NN scheme for the solution of TFFPMN model through mean square error based figure of merit, regression analysis, absolute error analysis and histograms. It is found that radial and azimuthal velocities are decaying functions of Deborah number. Further the nanofluid temperature enhances against higher radiation and thermophoresis parameters. The comparative assessment is performed to validate the present results.