Machine learning analysis of magnetic Oldroyd-B bioconvective nanofluid flow in biomedical systems.
Abstract
The succeeding demand for efficient thermal management and biomedical transport systems has strengthened interest in the complex flow behavior of biofluids under magnetized and bioconvective conditions. In this study, the nonlinear behavior of an Oldroyd-B Nanofluid (OBNF) in the presence of magnetism and biological convection formed by gyrotactic algae and Joule heating is examined. An artificial intelligence (AI) based computational framework has been developed in Python with the utilization of PyTorch to carry out the research. Using similarity transformations, the partial differential equations (PDEs) underlying the governing mechanics of the problem are converted into nonlinear ordinary differential equations (ODEs). A supervised multi-layer deep neural network (SM-DNN) structure has been constructed with Rectified Linear Units (ReLU) and Adam optimizers to forecast speed, temperature, concentration of nanoparticles, and density of microorganisms. Systematic research of the impact of various dimensionless numbers (magnetic number (M), bioconvection Rayleigh Number (Rb), mixed convection parameter (λ), Eckert number (Ec), Prandtl number (Pr), Lewis number (Le), Peclet number (Pe), and Biomass Lewis Number (Lb)) has been done. The SM-DNN has proved to have outstanding diagnosing capability with the R2 scores of more than 0.98 and mean squared errors (MSE) of less than 10-7. Based on the findings, it was noted that, in general, the increase of M and Rb led to a decrease of the pulsating velocity, while the occurrence of a higher Ec resulted in an increase of the flow temperature. Higher values of Le, Pr, Pe, and Lb were associated with a reduction of both concentration and motile density. The findings from this new methodology and artificial intelligence show great promise in the prediction of complex flow behavior of biofluids in medical science, thermal technology and transport engineering.