Scholarlink

<p>This study investigates the influence of thermal operating parameters on the hydrogen storage performance of La₀.₉Ce₀.₁Ni₅ and LaNi₄.₆Al₀.₄ intermetallic hydrides, while introducing a data-driven machine learning framework for predictive optimization. Hydrogen storage in metal hydrides is strongly governed by temperature and pressure, which significantly affect storage capacity as well as absorption–desorption kinetics, making system design inherently complex. To address this challenge, experimental data are leveraged to train and validate machine learning models, including AdaBoost and Random Forest, with a focus on identifying the most influential parameters for optimizing charging pressure and temperature. Among the models, Random Forest regression demonstrates superior predictive capability, achieving a coefficient of determination (R²) of 0.9998, mean absolute error (MAE) of 0.0014, root mean square error (RMSE) of 0.0031, and mean square error (MSE) of 0.0012. The close agreement between predicted and experimental results confirms the robustness and reliability of the proposed framework. Although the approach introduces increased computational complexity and training requirements, it significantly reduces the need for extensive experimental trials and detailed thermodynamic modeling. The findings establish an efficient pathway for determining optimal operating conditions in hydride-based hydrogen storage systems, enabling accelerated design and improved performance of next-generation energy storage technologies.</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

Predictive machine learning framework for optimizing hydrogen storage in La–Ni–Al hydrides

  • 期刊论文

  • 发表日期：

    2026年5月8日

查看全文
  • Kishor Kumar Sharma,
  • Asif Nizam,
  • Prem Kumar Chaurasiya,
  • Vinod Kumar Sharma,
  • Vikas Pandey

摘要

This study investigates the influence of thermal operating parameters on the hydrogen storage performance of La₀.₉Ce₀.₁Ni₅ and LaNi₄.₆Al₀.₄ intermetallic hydrides, while introducing a data-driven machine learning framework for predictive optimization. Hydrogen storage in metal hydrides is strongly governed by temperature and pressure, which significantly affect storage capacity as well as absorption–desorption kinetics, making system design inherently complex. To address this challenge, experimental data are leveraged to train and validate machine learning models, including AdaBoost and Random Forest, with a focus on identifying the most influential parameters for optimizing charging pressure and temperature. Among the models, Random Forest regression demonstrates superior predictive capability, achieving a coefficient of determination (R²) of 0.9998, mean absolute error (MAE) of 0.0014, root mean square error (RMSE) of 0.0031, and mean square error (MSE) of 0.0012. The close agreement between predicted and experimental results confirms the robustness and reliability of the proposed framework. Although the approach introduces increased computational complexity and training requirements, it significantly reduces the need for extensive experimental trials and detailed thermodynamic modeling. The findings establish an efficient pathway for determining optimal operating conditions in hydride-based hydrogen storage systems, enabling accelerated design and improved performance of next-generation energy storage technologies.