Artificial neural network approach to study the impact of gravity modulation on magneto-convection of Jeffrey fluid in a porous media

Abstract

The influence of gravity modulation on magneto-convection in Jeffrey fluid is analyzed. A weak nonlinear analysis is employed, utilizing a power series expansion method where disturbances are expressed as power series. Heat transfer is measured using the mean Nusselt number, $\overline{Nu}$. The study explores the application of machine learning techniques to predict the stability for different parameters. The Levenberg–Marquardt algorithm is used to train artificial neural networks on simulated data to understand the impact of magnetic fields on stability. The mean Nusselt number is estimated using the trained neural network. Regression $R$, histogram, and mean square error are used in the analysis. The developed artificial neural network model demonstrates its dependability because of its remarkable accuracy during the training, validation, and testing. The values of $R$ for Chandrasekhar number, $Q$, Jeffrey parameter, $\lambda$ and Magnetic Prandtl number, $P{r}_m$ are 99.93%, 94.22%, and 99.50% respectively. MSE curve is dropping, which is a indicator that the network is analyzing the training data and adjusting its weight to get the best fit. The values of best validation performance of Chandrasekhar number $Q$, Jeffrey parameter and Magnetic Prandtl number $P{r}_m$ are 3.6782E−06, 3.9345E−10, and 1.3218E−04 respectively.