Scholarlink

<p>Sustainable transportation is crucial in the global effort to reduce greenhouse gas emissions and transition to cleaner energy sources. In this context, hydrogen vehicles (HVs) have emerged as a promising and eco-friendly alternative to traditional fossil fuels. A key challenge in the advance of the hydrogen economy lies in the efficient distribution of hydrogen (HD), which is addressed through the hydrogen vehicle routing problem with time window and vehicle capacity (HVRPTC). In this paper, we explore the HVRPTC using HVs, taking into account factors such as driving costs, penalties for violating time windows, and vehicle capacities to determine optimal HD routes. While exact methods like the branch-and-bound (B&amp;B) algorithm offer robust solutions, their computational time can become prohibitive for larger instances. To solve this issue, we propose a hybrid solution by integrating B&amp;B and a deep learning (DL) model, where the B&amp;B method provides high-quality training instances and the trained DL models can predict near-optimal total cost with shorter computing time so that more effective solutions to HVRPTC can be developed in terms of both quality and efficiency. Our method promises encouraging results and provides solutions for a sustainable and efficient HD system. This study demonstrates the potential for scalable, real-time optimization of hydrogen logistics.</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

Integrating deep learning with branch-and-bound algorithm for enhanced solution of hydrogen distribution

  • 期刊论文

  • 发表日期：

    2025年12月1日

查看全文
  • Soukayna Abibou,
  • Dounia El Bourakadi,
  • Ali Yahyaouy,
  • Hamid Gualous

摘要

Sustainable transportation is crucial in the global effort to reduce greenhouse gas emissions and transition to cleaner energy sources. In this context, hydrogen vehicles (HVs) have emerged as a promising and eco-friendly alternative to traditional fossil fuels. A key challenge in the advance of the hydrogen economy lies in the efficient distribution of hydrogen (HD), which is addressed through the hydrogen vehicle routing problem with time window and vehicle capacity (HVRPTC). In this paper, we explore the HVRPTC using HVs, taking into account factors such as driving costs, penalties for violating time windows, and vehicle capacities to determine optimal HD routes. While exact methods like the branch-and-bound (B&B) algorithm offer robust solutions, their computational time can become prohibitive for larger instances. To solve this issue, we propose a hybrid solution by integrating B&B and a deep learning (DL) model, where the B&B method provides high-quality training instances and the trained DL models can predict near-optimal total cost with shorter computing time so that more effective solutions to HVRPTC can be developed in terms of both quality and efficiency. Our method promises encouraging results and provides solutions for a sustainable and efficient HD system. This study demonstrates the potential for scalable, real-time optimization of hydrogen logistics.