水力发电学报
            首 页   |   期刊介绍   |   编委会   |   投稿须知   |   下载中心   |   联系我们   |   学术规范   |   编辑部公告   |   English

水力发电学报 ›› 2025, Vol. 44 ›› Issue (8): 1-10.doi: 10.11660/slfdxb.20250801

• •    下一篇

特约论文:预报引导的水库调度综述

  

  • 出版日期:2025-08-25 发布日期:2025-08-25

Review of forecast informed reservoir operation

  • Online:2025-08-25 Published:2025-08-25

摘要: 水库作为重要的水资源工程,其高效运行具有重大意义。然而,国内外大多数水库基于历史统计资料制定调度规则,面向规划设计的静态控制限制了其主动应对洪水和干旱的能力。近年来,气象水文预报水平显著提高,预报引导的水库调度(FIRO)成为研究热点。本文介绍了现有的数值天气预报及其精度的主要影响因素、常用的水文预报模型以及快速发展的人工智能预报;综述了预报引导的水库调度方法及其应用成效。最后,建议综合利用包括全球导航卫星系统(GNSS)反演的大气可降水量在内的多种气象水文预报产品、制定面向“水汽—降水—径流”三道防线的预报调度方法、构建考虑梯级水库群多阻断效应的深度强化学习调度模型,以实现水资源的高效利用。

关键词: 水资源利用, 气象水文预报, 水库调度, 动态控制, 可降水量, 深度强化学习

Abstract: As key water resource projects, reservoirs and their efficient operation are crucial to the society. However, most reservoirs, domestic or international, have been formulating the operating rules based on historical statistical data, and the static rules for planning and design limit their capability of proactive responding to floods and droughts. Recent advancements in meteorological and hydrological forecasting have made forecast informed reservoir operation (FIRO) a research hotspot. This paper discusses the main factors that impact the accuracy of numerical weather forecasts, commonly used hydrological models, and rapidly developing artificial intelligence forecasting techniques. The review also covers the FIRO methods and their applications. Finally, we suggest the comprehensive use of various meteorological and hydrological forecasting products to achieve efficient water resource utilization-including global navigation satellite system (GNSS)-based precipitable water vapor, the development of FIRO methods focusing on the vapor-precipitation-runoff three lines of defense, and the construction of deep reinforcement learning operation models that account for the multi-blocking effects of cascade reservoirs.

Key words: water resource utilization, meteorological and hydrological forecast, reservoir operation, dynamic control, precipitable water vapor, deep reinforcement learning

京ICP备13015787号-3
版权所有 © 2013《水力发电学报》编辑部
编辑部地址:中国北京清华大学水电工程系 邮政编码:100084 电话:010-62783813
本系统由北京玛格泰克科技发展有限公司设计开发  技术支持:support@magtech.com.cn