Design of neural networks for Darcy–Forchheimer viscous fluid subjected to electro-magnetohydrodynamic and thermal impacts

Abstract

In this article, the electro-hydrodynamic Darcy–Forchheimer viscous fluid model EHDF−VFM over an extending surface is observed by operating the algorithm based on supervised learning through Artificial Levenberg Marquardt with backpropagated neural networks ALM−BPNNs. Energy correspondence is given the assistance of the primary law of thermodynamics in the presence of heat flux, dissipation, and Joule heating. Irreversibility investigation is demonstrated through the second law of thermodynamic. Partial differential equations of EHDF−VFM were transformed to ordinary differential equations using appropriate factors. These acquired non-linear ODEs are solved using the ND-solve procedure to take dataset of ALM−BPNNs for scenarios of this proposed model by varying E1, λ, Fr, Rd, Br, M, Pr, Sc, γ parameters and were graphically represented applying nftool to get performance, regression, fitness, error histogram, and training state analysis. Computational results for surface drag power, temperature slope, and mass exchange rate are discussed. Bejan number and entropy age have comparative effects on the electric field. For bigger porosity boundary, the entropy rate is expanded. An augmentation in porosity parameter decreases the velocity. The velocity profile shows conflicting conduct for higher approximation of electric and magnetic field parameters. The temperature has similar features for magnetic and radiation parameters.