Scholarlink

<p>This study addresses the pressing need for sustainable green hydrogen production in water-scarce regions, where traditional electrolysis-based methods are limited by freshwater availability. An integrated system was developed that combines thermoelectric condensation for atmospheric water harvesting with an alkaline water electrolyzer for hydrogen production. The system is powered by renewable solar energy, making it suitable for remote, arid environments. This study optimized system performance using three multilayer perceptron models to predict key output variables, including cell potential, hydrogen evolution rate, and current density, based on inputs such as temperature and electrode spacing. The training and prediction performance of network models trained using the Bayesian Regularization training algorithm has been analyzed comprehensively. Model results showed high predictive accuracy, with mean squared error values under 0.02, coefficient of determination values above 0.98 and average deviation rates as low as −&#xa0;0.002%. This research pioneered the application of thermoelectric condensation for green hydrogen production, distinguishing itself from previous work by integrating machine learning to fine-tune performance metrics. The findings suggested that machine learning-optimized atmospheric moisture capture could revolutionize sustainable hydrogen production, offering a scalable solution in water-limited 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

Predictive analytics for efficient green hydrogen production: machine learning applied to thermoelectric condensation-based atmospheric moisture capture

  • 期刊论文

  • 发表日期：

    2025年11月8日

查看全文
  • A. B. Çolak

摘要

This study addresses the pressing need for sustainable green hydrogen production in water-scarce regions, where traditional electrolysis-based methods are limited by freshwater availability. An integrated system was developed that combines thermoelectric condensation for atmospheric water harvesting with an alkaline water electrolyzer for hydrogen production. The system is powered by renewable solar energy, making it suitable for remote, arid environments. This study optimized system performance using three multilayer perceptron models to predict key output variables, including cell potential, hydrogen evolution rate, and current density, based on inputs such as temperature and electrode spacing. The training and prediction performance of network models trained using the Bayesian Regularization training algorithm has been analyzed comprehensively. Model results showed high predictive accuracy, with mean squared error values under 0.02, coefficient of determination values above 0.98 and average deviation rates as low as − 0.002%. This research pioneered the application of thermoelectric condensation for green hydrogen production, distinguishing itself from previous work by integrating machine learning to fine-tune performance metrics. The findings suggested that machine learning-optimized atmospheric moisture capture could revolutionize sustainable hydrogen production, offering a scalable solution in water-limited areas.