An intelligent framework for hybrid nanofluid flow over a curved stretching sheet via recurrent neural networks

In this article, the flow characteristics of a hybrid nanofluid containing AA7072 and AA7075 in water across a curved stretching surface are explored utilizing the non-Fourier model of heat flux. Hybrid nanofluids have diverse applications across biomedical, engineering, and industrial fields, including radiators, electronic cooling systems, and heat exchangers. This research examines the thermal performance of the proposed model considering constant wall temperature (CWT) and Newtonian heating (NH). In this study, a soft computing technique, specifically recurrent neural networks using the Levenberg–Marquardt optimizer (RNN-LMO), is presented to analyze the behavior of the non-Fourier model of heat flux. The behavior of key parameters affecting flow and temperature variations is analyzed through graphical representations and tabular descriptions. The Adam numerical technique is used to generate the dataset, and then RNN-LMO is applied to refine it. This combination provides approximate solutions of the proposed model presented in graphical and tabular form. The findings reveal that velocity increases with higher values of the curvature parameter. In the NH case, the temperature distribution is lower compared to the CWT case for various values of thermal relaxation and curvature parameters. A reduction in temperature distribution is noticed at higher values of the curvature and thermal relaxation parameters, but this reduction enhances the heat transfer rate. Additionally, the accuracy of the developed scheme is validated through the examination of error histograms, mean squared error curves, regression metrics, transition state, and absolute error analysis.