Machine learning-assisted modeling of mixed convection in tangent hyperbolic nanofluid with non-Fourier and non-Fickian transport

This investigation elaborates quadratic dual convection features in stretching sheet induced hyperbolic-tangent nanoliquid deploying modern fluxes based on generalized mass and heat transmission models. The modeled nonlinear transport expression accounts temperature-dependent conductivity, thermal generation and concentration-dependent diffusivity. Buongiorno nanomaterial diffusion model is modified in view of modern thermosolutal fluxes. The governing mathematical expressions derived subject to traditional boundary-layer concept are transfigured to ordinary differential framework by utilizing apposite similarity constraints. The numerical bvp4c approach is then utilized to solve the nonlinear equations, enabling the simulation to be precise. The intelligent ANNs-LMM model is trained and validated by utilizing reference dataset attained from the numerical Bvp4c approach with the help of MATLAB software, which proficiently handles the coupled nonlinear system of equations. Furthermore, the reference dataset is designed to cover the flow range of functional parameters scenarios, empowering all-inclusive testing, optimal validation and training performance of the neural network system. To show the various effects of functional parameters on the performance of proposed ANNs-LMM, histogram errors, mean square errors and regression plots are visualized for each case. The absolute errors between the reference dataset ranges from ${10}^{-8}$to${10}^{-10}$, reports the excellent accuracy of the intelligent ANNs-LMM approach. It is perceived that velocity profile escalates for increasing mixed convection parameter whereas it declines for higher values of power-law index parameter. The present findings are relevant in constructing nanoliquid-based systems in state-of-the-art manufacturing and coating technologies.