Scholarlink

<p>The present study develops an enhanced hybrid computational mechanics and machine learning framework for analyzing heat and mass transfer in skin tissue and smart materials. The methodology integrates the Cattaneo–Vernotte (CV) model, the Atangana–Baleanu (AB) fractional operator, and Laplace transform techniques to derive generalized thermoelastic formulations capable of capturing finite-speed thermal propagation, memory effects, and nonlocal stress relaxation. Closed-form expressions for temperature, displacement, dilation, and stress fields are obtained in the Laplace domain and numerically inverted to evaluate transient responses under thermal shock. All fractional thermoelastic simulations and Laplace inversions were executed in MATLAB R2023a, whereas the machine learning models (KNN, SVM, CNN) were implemented in Python 3.10 using scikit-learn and TensorFlow. To extend the predictive capacity of the analytical models, simulation-derived datasets are used to train three machine learning classifiers—K-nearest neighbors (KNN), support vector machine (SVM), and convolutional neural network (CNN). Comparative analyses through confusion matrices, dispersion maps, ROC curves, residual maps, and bar charts demonstrate that CNN achieves superior nonlinear feature extraction and generalization, SVM provides stable global decision boundaries, and KNN efficiently identifies localized thermal–mechanical anomalies. The AB fractional model is shown to suppress temperature overshoot and reduce stress concentration relative to CV, offering safer predictions for biological tissues. The combined fractional–ML framework enables rapid classification of safe and risky heating regimes, with potential applications in hyperthermia therapy, burn injury prevention, dermatological laser treatments, and thermal hotspot detection in engineered composites. This study establishes a unified pathway where fractional thermoelastic modeling, deep learning, and classical machine learning synergistically address complex biomedical and material thermal interactions. A synthetic dataset generated from fractional AB–CV thermoelastic simulations was used for training the ML classifiers.</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

Hybrid fractional thermoelastic–machine learning (KNN, CNN and SVM classifier) framework for heat and mass transfer: a computational mechanics approach

  • 期刊论文

  • 发表日期：

    2026年3月10日

查看全文
  • Jing Huang,
  • Amit Patil,
  • Abdulkafi Mohammed Saeed,
  • Sheela Hundekari,
  • Anjali Chaudhary,
  • Abhinav Singhal

摘要

The present study develops an enhanced hybrid computational mechanics and machine learning framework for analyzing heat and mass transfer in skin tissue and smart materials. The methodology integrates the Cattaneo–Vernotte (CV) model, the Atangana–Baleanu (AB) fractional operator, and Laplace transform techniques to derive generalized thermoelastic formulations capable of capturing finite-speed thermal propagation, memory effects, and nonlocal stress relaxation. Closed-form expressions for temperature, displacement, dilation, and stress fields are obtained in the Laplace domain and numerically inverted to evaluate transient responses under thermal shock. All fractional thermoelastic simulations and Laplace inversions were executed in MATLAB R2023a, whereas the machine learning models (KNN, SVM, CNN) were implemented in Python 3.10 using scikit-learn and TensorFlow. To extend the predictive capacity of the analytical models, simulation-derived datasets are used to train three machine learning classifiers—K-nearest neighbors (KNN), support vector machine (SVM), and convolutional neural network (CNN). Comparative analyses through confusion matrices, dispersion maps, ROC curves, residual maps, and bar charts demonstrate that CNN achieves superior nonlinear feature extraction and generalization, SVM provides stable global decision boundaries, and KNN efficiently identifies localized thermal–mechanical anomalies. The AB fractional model is shown to suppress temperature overshoot and reduce stress concentration relative to CV, offering safer predictions for biological tissues. The combined fractional–ML framework enables rapid classification of safe and risky heating regimes, with potential applications in hyperthermia therapy, burn injury prevention, dermatological laser treatments, and thermal hotspot detection in engineered composites. This study establishes a unified pathway where fractional thermoelastic modeling, deep learning, and classical machine learning synergistically address complex biomedical and material thermal interactions. A synthetic dataset generated from fractional AB–CV thermoelastic simulations was used for training the ML classifiers.