Deep learning analysis for enhanced prediction of heat transfer in Maxwell hybrid nanofluids with non-Fourier law and radiation effects

利用深度学习分析增强对具有非傅里叶定律和辐射效应的麦克斯韦混合纳米流体传热的预测。

阅读:1

Abstract

Nanofluids have emerged as promising heat transfer media due to their superior thermal performance associated to conventional fluids, yet their behavior under mutual impacts of radiation, heat source/sink, and magnetic fields remains an active area of research. In this study, movement and heat transfer characteristics of a Maxwell-based hybrid nanofluid containing carbon nanotubes are examined over a stretching/shrinking sheet with slip boundary conditions and Cattaneo–Christov heat flux. Governing nonlinear partial differential equations (PDEs) are reduced through similarity transformations and resolved numerically by means of MATLAB bvp4c routine. Numerical outcomes are analyzed for the effects of important parameters such as magnetic field strength, Biot number, velocity slip, and thermal radiation over velocity and temperature distributions. For the investigated parameter ranges (magnetic parameter 0.1–1.2, radiation parameter 0.3–0.9, and Biot number 0–0.8), thermal radiation and Biot number are found to enhance heat transfer rates, while increasing magnetic effects suppress the velocity field. The hybrid nanofluid consistently exhibits improved thermal performance compared to the corresponding nanofluid case. To complement numerical framework, artificial neural networks (ANNs) are implemented as surrogate predictive models. A multilayer perceptron structure qualified with Levenberg–Marquardt algorithm (LMA) is employed, with datasets from bvp4c distributed into 70% for training, 15% for validation, and 15% for testing. ANN demonstrates outstanding predictive performance, achieving mean squared error (MSE) values as low as [Formula: see text] for velocity gradients and [Formula: see text] for temperature profiles, with correlation coefficients of [Formula: see text] across all training, validation, and test subsets. These results confirm that ANN models can accurately reproduce the nonlinear boundary-layer solutions with negligible error. The combined numerical–ANN approach establishes a novel and computationally efficient framework for analyzing hybrid nanofluid flows. The outcomes highlight the strong influence of magnetic and thermal parameters on stream and heat transfer features and demonstrate ability of ANNs to serve as reliable predictive models for complex nonlinear systems.

特别声明

1、本页面内容包含部分的内容是基于公开信息的合理引用;引用内容仅为补充信息,不代表本站立场。

2、若认为本页面引用内容涉及侵权,请及时与本站联系,我们将第一时间处理。

3、其他媒体/个人如需使用本页面原创内容,需注明“来源:[生知库]”并获得授权;使用引用内容的,需自行联系原作者获得许可。

4、投稿及合作请联系:info@biocloudy.com。