Artificial neural network technique for simulation of improved thermal energy of Reiner–Philippoff nanofluid over a stretching curved surface

Abstract In mechanical engineering and industrial applications, single-walled carbon and multi-walled nanotubes are becoming more popular because of their direct impact on improving the thermal conductivity of base fluids. Considering the fascinating characteristics of carbon nanotubes (CNTs), the objective of this study is to examine the influence of radiation on the movement of nanofluids suspended in CNTs across a stretched sheet affected by slip state, while a magnetic field is present. Combining neural networks with the backpropagation technique based on the Levenberg–Marquardt scheme is a new computer model that this study proposes. Applying it to the flow of non-Newtonian Riner–Philippoff boundary layers via a curved tensile plate allows one to derive non-linear systems from the governing equations and solve them. Furthermore, the partial differential equations for the magnetohydrodynamics boundary layer flow across a curved stretched sheet are transformed into non-similar dimensionless partial differential equations, which are then handled as ordinary differential equations using the local non-similarity method and solved using the bvp4c MATLAB tools. By adjusting the following embedding settings, a variety of scenarios are used to generate an input data set for the presented Levenberg Marquardt scheme- Back-Propagation Neural Network (LMS-BPNN) model: Bingham, Riener–Philippoff fluid, and magnetic parameter. To compare the produced scenarios' outcomes with the reference results, we assess the LMS-BPNN model's training, testing, and validation. The efficacy and performance of the LMS-BPNN infrastructure model are assessed using a variety of metrics for the fluidic system convergence study, among which are regression graphs, an error histogram, and the mean square error (MSE). Using the built-in scenarios, we evaluate the training, testing, and validation of the LMS-BPNN model while assessing the findings to the standard outcomes. MSE, error histogram, and regression plots are used to evaluate the LMS-BPNN infrastructure model for the fluidic system convergence study.