Artificial intelligence-based procedure to analyze heat transfer features for chemically reactive Darcy-Forchheimer flow of magnetized tetra-hybrid nanofluid flow capturing Joule heating aspects through stenotic artery

Abstract

Scientists are studying the thermal characteristics and energy requirements of blood containing magnetic nanoparticles using the Darcy-Forchheimer flow model (DFFM) framework. This research examines various heat transfer mechanisms, including thermal radiation and viscous effects, to advance medical treatments by understanding how these particles behave in arterial blood flow under magnetic fields. The study employs a tetra hybrid nano-fluid consisting of silver ($Ag$), gold ($\mathrm{Au}$) and titanium-oxide $\left(Ti{O}_2\right)$, copper $\left(Cu\right)$, and blood is treated as the base fluid. Gold, copper, and silver nano-particles are taken for their potential in imaging and drug delivery. The objective of this study is to evaluate the performance of tetra hybrid nano-fluid models (THNFM) using numerical approach along with AI integrated analysis. Using artificial intelligence techniques that incorporate predicting methods yields accurate predictions for this complex fluid system. The model accounts for both random variables and turbulent flow characteristics. Mathematical simplification is achieved by converting the governing partial differential equations to ordinary differential equations through similarity methods. ND-Solver Mathematica technique is employed to solve converted ODEs to gain behavior of different parameters. After getting tabular data MATLAB tool is used to process it for Neural Fitting mechanism. The dataset is split into 70 %, 15 %, and 15 % for training, testing, and validation in the Neural Network-Back propagation Levenberg-Marquardt scheme (NN-BPLMS). NN-BPLMS is applied to draw graphs of performance, state function, error histogram, regression analysis and function fit numerically using the sequence of train, validate and test (TTV) in NN-BPLMS processes. Applied technique exhibited the approximate solutions of TETHMNF DFFM for different cases and comparison with reference results to verify the correctness of the proposed NN-BPLMS. This applied Neural Network techniques effectively solved the TETHMNF performance with squared mean error (MSE), reflections of regression analysis (RA) and histogram studies (EHA). The comparison between the results obtained prior and after implication of the AI-NNs, the errors ranging from ${10}^{-7}$ to${10}^{-5}$ are observed, which confirms the accuracy of the method.