Scholarlink

<p>The study of micropolar fluid flow in exponentially curved surfaces has profound implications and improves understanding of complex fluid dynamics to increase theoretical knowledge and predictive modeling for practical uses in manufacturing, thermal control, and biomedical engineering. The current research probes the flow behavior of micropolar fluids on the exponentially curved surface (MPFs-ECS) using a celebrated artificial intelligence-based nonlinear autoregressive exogenous deep learning network (NARXDLN) trained by the Levenberg–Marquardt algorithm (LMA), i.e., NARXDLN-LMA. The synthetic dataset is created through the implementation of the Lobatto IIIA technique to determine the effect of variations in the Prandtl and Eckert numbers, magnetic and material parameters, and radius of curvature on the temperature, axial velocity distribution, and microrotation velocity profiles to train the neurocomputational architecture. The NARXDLN-LMA is applied to the achieved datasets to simulate and study the convoluted dynamics of the MPFs-ECS model by segmenting into testing, training, and validation subsets. The proposed architecture is extensively tested by the performance analysis of the mean square errors at testing, training, and validation stages, while further assessments are made through the error auto-correlation, histogram errors, and regression assessments between exogenous variables and error. The findings of the comparative evaluation of the FPFs-ECS model show that the NARXDLN-LMA can effectively capture the dynamic behavior, being very accurate with error values between 10<sup>–02</sup> and 10<sup>–10</sup>, trustworthy for sundry scenarios on the basis of sufficient experimentation.</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

A Novel Design of Labatto IIIA Data-Driven Intelligent Exogenous Deep Neurocomputing Architecture for Thermophysics Analysis of Micropolar Fluid Flow Model with Exponentially Curved Surface

  • 期刊论文

  • 发表日期：

    2026年4月13日

查看全文
  • Nabeela Anwar,
  • Arfa,
  • Muhammad Shoaib,
  • Adiqa Kausar Kiani,
  • Muhammad Asif Zahoor Raja

摘要

The study of micropolar fluid flow in exponentially curved surfaces has profound implications and improves understanding of complex fluid dynamics to increase theoretical knowledge and predictive modeling for practical uses in manufacturing, thermal control, and biomedical engineering. The current research probes the flow behavior of micropolar fluids on the exponentially curved surface (MPFs-ECS) using a celebrated artificial intelligence-based nonlinear autoregressive exogenous deep learning network (NARXDLN) trained by the Levenberg–Marquardt algorithm (LMA), i.e., NARXDLN-LMA. The synthetic dataset is created through the implementation of the Lobatto IIIA technique to determine the effect of variations in the Prandtl and Eckert numbers, magnetic and material parameters, and radius of curvature on the temperature, axial velocity distribution, and microrotation velocity profiles to train the neurocomputational architecture. The NARXDLN-LMA is applied to the achieved datasets to simulate and study the convoluted dynamics of the MPFs-ECS model by segmenting into testing, training, and validation subsets. The proposed architecture is extensively tested by the performance analysis of the mean square errors at testing, training, and validation stages, while further assessments are made through the error auto-correlation, histogram errors, and regression assessments between exogenous variables and error. The findings of the comparative evaluation of the FPFs-ECS model show that the NARXDLN-LMA can effectively capture the dynamic behavior, being very accurate with error values between 10–02 and 10–10, trustworthy for sundry scenarios on the basis of sufficient experimentation.