Scholarlink

<p>Large-scale open-channel irrigation systems in China are characterized by multilevel (five or more levels) canals with a tree topology, incorporating hundreds to thousands of gates. Effective management necessitates integrated water level regulation and distribution across pools to ensure operational efficiency, safety, and fulfillment of water demands. The complexity of these systems presents formidable challenges for modeling and optimization. Current frameworks for open-channel irrigation systems do not adequately account for the coupled dynamics of water level regulation and water distribution, which introduces considerable uncertainty and leads to suboptimal or infeasible solutions. This study introduces a graph-based mixed-integer linear programming framework that integrates these three components. By a novel sparsity-guided mechanism rooted in extreme direction theory, it strengthens the coupling between gate operations and water level fluctuations, optimizing system performance to meet practical engineering demands in real-time. The model’s performance in safety and efficiency was assessed through nine real-world networks of varying complexity. Results demonstrate that an optimized problem structure improves computational efficiency, reducing linear programming (LP) solution time by 14% and mixed-integer linear programming (MILP) solution time by 35.7%. In irrigation scheduling, the model reduced the irrigation duration by 18.9%. Beyond facilitating water network optimization, the model effectively captures the coupled dynamics of check structures and offtakes, strengthening its utility in large-scale open-channel irrigation network.</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

OcisMILPNet: A Graph-Based MILP Approach for Sparse Operations in Open-Channel Irrigation Networks

  • 期刊论文

  • 发表日期：

    2026年7月21日

查看全文
  • Wentao Hou,
  • Haorui Chen,
  • Shaohui Zhang,
  • Jing Fu,
  • Juyan Mu,
  • Boxiong Zhang,
  • Meijian Bai,
  • Yinong Li,
  • Baozhong Zhang

摘要

Large-scale open-channel irrigation systems in China are characterized by multilevel (five or more levels) canals with a tree topology, incorporating hundreds to thousands of gates. Effective management necessitates integrated water level regulation and distribution across pools to ensure operational efficiency, safety, and fulfillment of water demands. The complexity of these systems presents formidable challenges for modeling and optimization. Current frameworks for open-channel irrigation systems do not adequately account for the coupled dynamics of water level regulation and water distribution, which introduces considerable uncertainty and leads to suboptimal or infeasible solutions. This study introduces a graph-based mixed-integer linear programming framework that integrates these three components. By a novel sparsity-guided mechanism rooted in extreme direction theory, it strengthens the coupling between gate operations and water level fluctuations, optimizing system performance to meet practical engineering demands in real-time. The model’s performance in safety and efficiency was assessed through nine real-world networks of varying complexity. Results demonstrate that an optimized problem structure improves computational efficiency, reducing linear programming (LP) solution time by 14% and mixed-integer linear programming (MILP) solution time by 35.7%. In irrigation scheduling, the model reduced the irrigation duration by 18.9%. Beyond facilitating water network optimization, the model effectively captures the coupled dynamics of check structures and offtakes, strengthening its utility in large-scale open-channel irrigation network.