AI driven RNN approach for investigation of thermal features in magneto-radiative nanofluid under random microbial movement

Abstract

Recurrent neural network (RNN) applications in fluid dynamics have transformed the field by making it possible to model complex fluid behaviors with previously unattainable accuracy, thereby significantly improving the ability to forecast. Recurrent Neural Networks (RNN) has been employed as an AI tool to study thermal radiation in the MHD flow of gyrotactic organisms with nanoparticles and velocity slips. This investigation reports the bioconvective-MHD inspired flow of Casson fluid under certain physical effects. The model discussed through RNN approach. Different scenarios are examined to investigate how convergence parameters affect chemical reactions and heat generation/absorption. The significant results for thermal radiations, MHD and slip effects are analyzed. The Bvp4c approach is used to solve the transformed ODEs computationally. The synthetic datasets are obtained using mathematical simulation of the bvp4c numerical approach for TR-MHD-FGONV. Then, the supervised computing RNN approach is applied to the synthetic datasets for every model variant; the RNN findings exhibit tiny errors and closely match numerical observations. The effectuality of RNNs is meticulously proven through holistic experiments, validating iterative convergence rates for mean squared error (MSE), optimization controlling measurements, and error distribution using histograms. The fundamental consequence illustrates the contribution of the various parameters to the fluid's flow.