Scholarlink

<p>This study presents a data-driven investigation of energy and exergy performance of parabolic trough solar collectors (PTSCs) using ensemble machine learning regression models. To predict the energy and exergy efficiencies, a comprehensive dataset comprising thermo-fluid and solar operating parameters such as Reynolds number, nanoparticles volume fraction, inlet fluid temperature, and direct solar irradiance is used. The dataset considered three heat transfer fluids, “Dowtherm Q”, “Syltherm 800”, and “Therminol VP-1”, each blended with nanoparticles “Al₂O₃”, “CuO”, and “SiO₂”. Three ensemble algorithms namely AdaBoost, Gradient Boosting, and XGBoost, were trained and optimised through random search hyperparameter tuning and validated using five-fold cross-validation. The model’s performance was evaluated using mean squared error (<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(\:MSE\)</EquationSource> </InlineEquation>), mean absolute error (<InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(\:MAE\)</EquationSource> </InlineEquation>), and coefficient of determination (<InlineEquation ID="IEq3"> <EquationSource Format="TEX">\(\:{R}^{2}\)</EquationSource> </InlineEquation>). The results show that Gradient Boosting had the highest accuracy (<InlineEquation ID="IEq4"> <EquationSource Format="TEX">\(\:{R}^{2}&gt;0.99\)</EquationSource> </InlineEquation> for energy and up to <InlineEquation ID="IEq5"> <EquationSource Format="TEX">\(\:0.9999\)</EquationSource> </InlineEquation> for some cases) but AdaBoost also performed well (<InlineEquation ID="IEq6"> <EquationSource Format="TEX">\(\:{R}^{2}&gt;0.97\)</EquationSource> </InlineEquation> for exergy). And XGBoost underperformed (<InlineEquation ID="IEq7"> <EquationSource Format="TEX">\(\:{R}^{2}&lt;0.53\)</EquationSource> </InlineEquation>), confirming boosting-based ensembles are the most reliable models for PTSC efficiency prediction. This work provided a novel framework for integrating data-driven models into performance analysis and optimization of solar thermal systems.</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

Ensemble learning-based prediction of energy and exergy performance in nanofluid-driven parabolic trough solar collectors

  • 期刊论文

  • 发表日期：

    2026年5月13日

查看全文
  • Himanshu Upreti,
  • Ankita Pandey,
  • Manisha Saini,
  • Ziya Uddin

摘要

This study presents a data-driven investigation of energy and exergy performance of parabolic trough solar collectors (PTSCs) using ensemble machine learning regression models. To predict the energy and exergy efficiencies, a comprehensive dataset comprising thermo-fluid and solar operating parameters such as Reynolds number, nanoparticles volume fraction, inlet fluid temperature, and direct solar irradiance is used. The dataset considered three heat transfer fluids, “Dowtherm Q”, “Syltherm 800”, and “Therminol VP-1”, each blended with nanoparticles “Al₂O₃”, “CuO”, and “SiO₂”. Three ensemble algorithms namely AdaBoost, Gradient Boosting, and XGBoost, were trained and optimised through random search hyperparameter tuning and validated using five-fold cross-validation. The model’s performance was evaluated using mean squared error ( \(\:MSE\) ), mean absolute error ( \(\:MAE\) ), and coefficient of determination ( \(\:{R}^{2}\) ). The results show that Gradient Boosting had the highest accuracy ( \(\:{R}^{2}>0.99\) for energy and up to \(\:0.9999\) for some cases) but AdaBoost also performed well ( \(\:{R}^{2}>0.97\) for exergy). And XGBoost underperformed ( \(\:{R}^{2}<0.53\) ), confirming boosting-based ensembles are the most reliable models for PTSC efficiency prediction. This work provided a novel framework for integrating data-driven models into performance analysis and optimization of solar thermal systems.