Scholarlink

<p>With changing global climate conditions, drought has emerged as one of the most devastating natural disasters affecting the reliability of the hydropower supply. In response to these extreme drought events, this study proposes an AI-driven prediction-decision framework to increase the accuracy of predicting potential droughts and optimizing scheduling for cascaded hydropower–hydrogen systems. This study uses the Pearson-III distribution to categorize historical runoff patterns and identify extremely dry years for analysis. With limited available runoff data, the DoppelGANger model and eXtreme Gradient Boosting (XGBoost) are employed to generate scenarios and forecast future daily runoff based on runoff data from these extremely dry years. A joint scheduling model for hydropower–hydrogen systems is then developed to assess the effect of hydrogen on improving power supply capabilities with restricted generation flow. On the basis of historical runoff data from the Heishui River in Sichuan Province, the integration of hydrogen fuel cells into cascade hydropower stations has resulted in a significant improvement in system resilience of 65.09%. A comprehensive analysis is conducted to evaluate the economic benefits of the system, considering factors such as the installed capacity of hydrogen fuel cells, runoff levels, and proportion of hydropower in the power generation structure.</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 data-driven prediction-decision framework for improving the resilience of hydropower systems under drought disasters

  • 期刊论文

  • 发表日期：

    2025年12月17日

查看全文
  • Jingsi Huang,
  • Haowei Shao,
  • Feng Gao,
  • Jie Song,
  • Guannan He

摘要

With changing global climate conditions, drought has emerged as one of the most devastating natural disasters affecting the reliability of the hydropower supply. In response to these extreme drought events, this study proposes an AI-driven prediction-decision framework to increase the accuracy of predicting potential droughts and optimizing scheduling for cascaded hydropower–hydrogen systems. This study uses the Pearson-III distribution to categorize historical runoff patterns and identify extremely dry years for analysis. With limited available runoff data, the DoppelGANger model and eXtreme Gradient Boosting (XGBoost) are employed to generate scenarios and forecast future daily runoff based on runoff data from these extremely dry years. A joint scheduling model for hydropower–hydrogen systems is then developed to assess the effect of hydrogen on improving power supply capabilities with restricted generation flow. On the basis of historical runoff data from the Heishui River in Sichuan Province, the integration of hydrogen fuel cells into cascade hydropower stations has resulted in a significant improvement in system resilience of 65.09%. A comprehensive analysis is conducted to evaluate the economic benefits of the system, considering factors such as the installed capacity of hydrogen fuel cells, runoff levels, and proportion of hydropower in the power generation structure.