Machine Learning-Based Simulation and Analysis of MHD Non-Newtonian Fluid Dynamics in Blade Coating Processes

The complex dynamics of magnetohydrodynamic (MHD) non-Newtonian fluid flow in the blade coating process using Brownian, thermophoresis, and magnetic effects are examined in this research which is important in surface engineering. The nonlinearities and parameter sensitivities present in these systems are frequently difficult for traditional numerical techniques to handle. To overcome these obstacles, we present a machine learning (ML) based framework using artificial neural networks (ANNs), to simulate and analyze the behavior of non-Newtonian fluids under the influence of magnetic fields. A hybrid method that combines numerical simulation and ANN-based surrogate modelling is used to solve the governing partial differential equations. Lubrication approximation theory (LAT) simplifies mathematical equations derived from mass, momentum, and energy conservation laws. The numerical algorithm called shooting technique is used to solve the dimensionless equations. A comparative analysis of numerical data with ANN revealed highly similar findings. The suggested model, supervised neural networks with Levenberg-Marquardt backpropagation (LMBPSNNs) framework, obtains approximation solutions for different flow conditions. It is determined whether the suggested LMBPSNN solver is effective by doing comparative analyses. Also, the effects of the slip parameter, and MHD on coating thickness, blade velocity, temperature, and pressure, are studied in detail. Non-Newtonian parameters and Hartmann number are the controlling parameters in determining the thin layer coating thickness, leading to better web life and efficient coating process. The study demonstrates how combining fluid dynamics and ML can improve predictive modelling, shorten simulation times, and facilitate intelligent process design in intricate MHD systems.