Critical points and stability analysis in MHD radiative non-Newtonian nanoliquid transport phenomena with artificial neural network prediction

摘要

In the present framework, flow and thermal transport behavior of non-Newtonian viscoelastic fluid induced by stretching/shrinking of the horizontal sheet under the influence of Lorentz force, volumetric heat source/sink, and radiation (assuming optically thick medium) has been investigated. Multiple solutions (Branches) have been predicted numerically using Lie symmetry transformation and Runge Kutta Dormand Prince (RKDP) algorithm-based Shooting method for the different controlling parameters, stretching/shrinking (gamma c 0. Some of the results have also been compared with MATLAB built-in solvers to validate our in-house code. The deviations in critical (turning) points (beta(c),gamma(c)) have been noticed for different values of M and alpha. The temporal stability analysis is performed for these parameters (beta,gamma), and the upper (lower) branch is only physically feasible (infeasible). The combined effect on Nusselt number and skin-friction coefficient is also depicted in the form of contour plots. The outcomes from artificial neural network (ANN) using Levenberg-Marquardt algorithm give a preferred estimate for current numerical simulation. In ANN prediction, the optimal number of neurons in the hidden layer is used, having minimum value of RMSE and RMRE with maximum R-2 for different cases of non Newtonian stretching/shrinking sheet. Finally, the results conclude that the applied ANN model can accurately predict the output results.