Artificial neural networks analysis for non-newtonian nanofluid flow with variable viscosity and MHD effects in wire covering processes

Abstract

Nanofluids enhance the thermal conductivity and heat transmission rate of the base fluid, reducing the risk of overhating which increase the lifespan of the coating wires. In the realm of artificial neural networks, the Levenberg-Marquardt Algorithm is characterized by its innovative stability and produces a computational resolution of the wire covering for third grade nanofluid flow (WC-TGNFF) utilizing regression plots (RP), histogram visualizations, state transition metrics, and mean squared errors (MSE). This manuscript examines WC-TGNFF through the implementation of a novel intelligent computing system via the Levenberg-Marquardt Algorithm (ANN-LMA). The basic flow equations in expressions of PDEs for WCS-TGNFF is transformed into non-dimensional ODEs. The data acquisition for the proposed ANN-LMA is generated for parameters involved in the model WCS-TGNFF through Runge-Kutta method. The training, validation, and testing phases of ANN-LMA are employed to assess the results derived from WC-TGNFF across diverse scenarios, and a comparative analysis of the derived results is conducted against a reference dataset to evaluate the precision and efficacy of the proposed ANN-LMA framework in addressing non-Newtonian fluid challenges associated with WC-TGNFF. The novel contribution of the present model is to investigate the Brownian and thermophesis effects via ANN-LMA on the wire coating. The remarkable agreement of the proposed findings with reference solutions underscores the robustness of the framework, achieving a precision level 10−6.

Keywords

• Wire coating
• Third grade nanofluid model
• Artificial neural network
• Levenberg-marquardt algorithm
• Variable viscosity