Scholarlink

<p>The current study presents a comparative analysis to validate the computing efficiency of artificial neural network (ANN) with Levenberg–Marquardt feed-forward backpropagation algorithm (LMFFBPA) against numerical method (NM) and multiple linear regression (MLR) for analysing the optimization of entropy in MHD nanofluid convective flow over two distinct geometries. The impact of nonlinear radiation, Brownian motion, Cattaneo–Christov heat and mass fluxes, and thermophoretic particle deposition under the convective boundary conditions are taken into consideration. By applying the appropriate transformations, the system of nonlinear PDEs (Partial differential equations) is converted into a set of nonlinear ODEs. The resulting nonlinear system of ODEs (Ordinary differential equations) is solved using NM. The acquired datasets are utilised to train, validate and test LMFFBPA, enabling precise predictions of the skin friction coefficient, Nusselt number and Sherwood number. Examination of mean square error (MSE), error histogram, and regression fitness confirms the proficiency of ANN. These evaluations ensure the accuracy and reliability of the present outcomes. The regression at the training, validation, testing, and all stages tends to be almost 1, indicating optimal ANN performance with high accuracy and minimal error. Based on the comparisons between ANN and MLR, it can be determined that the ANN exhibits superior performance with a high level of accuracy. Artificial neural networks using the Levenberg–Marquardt technique improve simulation speed and precision, hence facilitating improved design and optimization. This research will yield practical real-world solutions in material design and manufacturing, energy generation and conversion, thermal management systems, and other related areas.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Levenberg–Marquardt neural network analysis of entropy optimization on MHD nanofluid convective flow with nonlinear thermal radiation and Cattaneo–Christov heat and mass fluxes: a comparative study

  • 期刊论文

  • 发表日期：

    2025年5月30日

查看全文
  • M. Priya,
  • P. Bala Anki Reddy

摘要

The current study presents a comparative analysis to validate the computing efficiency of artificial neural network (ANN) with Levenberg–Marquardt feed-forward backpropagation algorithm (LMFFBPA) against numerical method (NM) and multiple linear regression (MLR) for analysing the optimization of entropy in MHD nanofluid convective flow over two distinct geometries. The impact of nonlinear radiation, Brownian motion, Cattaneo–Christov heat and mass fluxes, and thermophoretic particle deposition under the convective boundary conditions are taken into consideration. By applying the appropriate transformations, the system of nonlinear PDEs (Partial differential equations) is converted into a set of nonlinear ODEs. The resulting nonlinear system of ODEs (Ordinary differential equations) is solved using NM. The acquired datasets are utilised to train, validate and test LMFFBPA, enabling precise predictions of the skin friction coefficient, Nusselt number and Sherwood number. Examination of mean square error (MSE), error histogram, and regression fitness confirms the proficiency of ANN. These evaluations ensure the accuracy and reliability of the present outcomes. The regression at the training, validation, testing, and all stages tends to be almost 1, indicating optimal ANN performance with high accuracy and minimal error. Based on the comparisons between ANN and MLR, it can be determined that the ANN exhibits superior performance with a high level of accuracy. Artificial neural networks using the Levenberg–Marquardt technique improve simulation speed and precision, hence facilitating improved design and optimization. This research will yield practical real-world solutions in material design and manufacturing, energy generation and conversion, thermal management systems, and other related areas.