Artificial neural network model using Levenberg Marquardt algorithm to analyse transient flow and thermal characteristics of micropolar nanofluid in a microchannel

Abstract

The application of artificial neural network has kept the whole world amazed, as it has flourished its roots to analyse different domains of science and technology. The current article is modelled to bear witness to how the artificial neural network is administered to study heat transfer and fluid flow problems. The model constructed analyses the mixed convective and unsteady flow of micropolar fluid through the microchannel in the presence of activation energy and magnetic field using Buongiorno's model. No slip and convective boundary conditions are employed. The partial differential equation is solved using the finite difference method. The artificial neural network using the Levenberg-Marquardt algorithm with the feed-forward backpropagation method is constructed and trained. The results of the analysis show that the material parameter lowers the fluid's velocity. For higher magnetic effects, the micro-rotation profile maximises at left half and minimises at right half of microchannel. The temperature profile increases with increasing Eckert number and thermophoresis parameter. The reaction rate parameter is a depleting function, while the activation energy parameter is an enhancing function of the solutal profile. The results obtained from the artificial neural network for all 8 scenarios are highly reliable due to its high accuracy, which is pleasantly deliberated by the mean square error values, error histograms, training, and regression graphs of the neural network model. The absolute error analysis carried out is in the range of 10⁻⁴ to 10⁻⁵. The prominent conclusion from the analysis is that artificial neural network is sophisticated tool to predict the subsequent sequel of fluid flow and heat transport over a long period of time, reducing computational time to solve complicated fluid flow problems.