Levenberg–Marquardt neural network for entropy optimization on Casson hybrid nanofluid flow with nonlinear thermal radiation: a comparative study

The purpose of this study is to investigate entropy optimization in the magneto-hydrodynamic and electro-magneto-hydrodynamic flow of a Casson hybrid nanofluid over a rotating disk with nonlinear thermal radiation. The governing dimensional partial differential equations were reduced to ordinary differential equations by using appropriate transforms and solved numerically. The effects of several physical factors on the velocity, temperature, entropy generation, Bejan number, Nusselt number, and skin friction coefficient in comparison to the nanofluid and hybrid nanofluid scenarios over a rotating disk are explored both tabularly and graphically. The constructed artificial neural network is the most appropriate for predicting the skin friction coefficient and Nusselt number over a rotating disk. As the magnetic field strength increased, the velocity profiles decreased in the nanofluid and hybrid nanofluid scenarios. When the thermal radiation increased, the amount of entropy generated for the nanofluids and hybrid nanofluids also increased. We built the artificial neural networking model using 51 sample values of the skin friction coefficient and Nusselt number as outputs. This section provides various dimensionless parameters, which are all inputs. We utilized 70% of the data for training, and 15% for validation and testing. The Levenberg–Marquardt algorithm and back-propagation were used to train the neural network. The best validation performance for skin friction and the Nusselt number for the Casson hybrid nanofluid across a rotating disk are 6652e-07 at epoch 138 and 2.7094e-05 at epoch 7. Additionally, the training, validation, testing, and performance of the ANN model were closer to unity.