Scholarlink

This paper addresses the challenge of short-term multi-reservoir hydropower optimization, comparing Reinforcement Learning (RL) models with a benchmark Mixed-Integer Linear Programming (MILP) approach. While existing literature focuses on long-term objectives, this study explores short-term decision-making crucial for daily operations. Three RL algorithms are tested across different environment configurations. Results demonstrate RL approaches the MILP’s performance while requiring much less computational time.

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

Multi-reservoir Hydropower Optimization with Alternative Deep Reinforcement Learning Algorithms

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Rodrigo Castro-Freibott,
  • Alvaro García-Sánchez,
  • Francisco Espiga-Fernández,
  • Guillermo González-Santander

摘要

This paper addresses the challenge of short-term multi-reservoir hydropower optimization, comparing Reinforcement Learning (RL) models with a benchmark Mixed-Integer Linear Programming (MILP) approach. While existing literature focuses on long-term objectives, this study explores short-term decision-making crucial for daily operations. Three RL algorithms are tested across different environment configurations. Results demonstrate RL approaches the MILP’s performance while requiring much less computational time.