Artificial neural network–assisted comparative analysis of variable viscosity models in polymer wire coating

Purpose An ANN-based computational framework is designed to examine the role of variable viscosity in the wire-coating process. For this, two well-known viscosity models, namely the Vogel and Arrhenius, are considered. Third-grade fluid flow in the presence of radiative heat, variable viscosity and a transverse magnetic field is measured. A comparative study is performed to understand the role of each model on the flow and thermal outlines in the coating region. Design/methodology/approach The modeled nonlinear differential equations were numerically solved using the MATLAB bvp5c function, and the ANN–LMA concept was applied to analyze regression, mean-squared error, state transition dynamics and error histogram results. Findings A parametric analysis is performed to determine the significance of each parameter in fiber coating. The Vogel’s model predicts relatively stable flow behavior under varying parameters compared with the Arrhenius model. The artificial neural network (ANN) performance, training state, error histogram and regression analysis reveal fast convergence, low mean squared error and a perfect regression coefficient (R = 1), demonstrating the precision and reliability of the proposed model. Originality/value Published works have not analyzed the role of variable viscosity in the wire-coating process. This study fills the above gap through a comparative analysis using ANNs, considering third-grade fluid. This study also aims to identify the optimal viscosity model from the Vogel and Arrhenius models.