Applications of neural networking in Eyring-Powell nanofluid dynamics on a rotating surface in a porous medium

One of the fundamental aspects of solving difficult and nonlinear mathematical ideas is the use of Artificial Neural Networks due to their exceptional efficiency in handling such problems. In many complex fields such as computational fluid system, biological computation, and biotechnology, a distinct computing structure is provided by Artificial Neural Networks, which is extremely valuable. The main purpose of this article is to dig out the abilities of the Levenberg-Marquardt technique using back-propagation artificial neural networks regarding the fluid mechanics and the heat transport assessment in nanoparticles. This interdisciplinary field explores heat and mass transfer through objects and fluids, and their impacts on temperature as well as concentration distributions. With the help of mathematical modelling and numerical solution methodologies, researchers can simulate and analyze these processes. The present analysis communicates the Eyring-Powell fluid flow caused by a rotating disk placed in the horizontal direction. The fluid flow over a rotating disk through non-linear partial differential equations is modeled. After converting the partial differential equations to ordinary ones, they are tackled numerically through shooting technique. The Levenberg-Marquardt algorithm using back-propagation artificial neural networks technique is used with reference datasets, having 70 % training, 15 % testing, and 15 % to validation. The method is validated with the help of mean squared error, error histogram and comprehensive regression analysis. These figures show the accuracy of the proposed method for solving nonlinear problems. Flow features such as velocity, temperature as well as the concentration profiles are exemplified quantitatively and have been graphically discussed. Velocity of the fluid decreases for porosity and increases for fluid parameter while temperature increases for thermophoresis and Brownian motion parameters. Consistency is shown by getting a minimum absolute error approaching zero, showing the strength of the proposed approach.