A PINN framework for the prediction of thermo-viscous flows between extended flat plates embedded in a porous slab with suction/injection

A powerful framework Physics-Informed Neural Networks (PINNs) provides solutions of complicated engineering challenges by incorporating governing physical theories with noisy or inadequate data. The present study relates PINNs to predict thermo-viscous flows between extended flat plates embedded in a porous slab with suction and injection. The PINNs optimizes biases and weights to estimate the results of required modelled equations by incorporating physical limitations into the loss function networks. The key parameters such as the thermal stress, strain conductivity, Prandtl number, medium porosity and suction/injection are examined. The influence of all these factors on flow velocity and temperature fields have been examined with the help of graphical interpretations. The precision of the PINN-based methodology is further validated via. Comparative investigation with Finite Difference Method(FDM) solver developed in Python. The various statistical measures supporting strong agreement and strengthening its efficiency in solving current investigated equations.