Scholarlink

<p>The appropriate design of an energy management system (EMS) is crucial in fuel cell hybrid electric vehicles (FCHEVs) as it must ensure an efficient coordination of the multiple energy sources including the fuel cell (FC), battery, and super-capacitor (SC). This paper presents an up-to-date and extensive review of EMS strategies, focusing on RL-based methodologies and their application on new powertrain architectures. RL is identified as a promising technology that could be used to implement intelligent and adaptive control solutions that would also be energy efficient for FCHEVs. The review focuses on four main aspects: (1) powertrain configurations adopted for FCHEVs, (2) reinforcement learning techniques, parameters, and performance measures, (3) the use of multi-objective optimization for EMS design, and (4) datasets and simulation cycles applied in the studies being reviewed. Special focus is placed on the ability of the RL framework to learn dealing with conflicting objectives. These insights aim to guide researchers in advancing next-generation EMS for FCHEVs.</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

Reinforcement learning for energy management in fuel cell hybrid electric vehicles: an up-to-date review with topology-centric insights

  • 期刊论文

  • 发表日期：

    2026年1月31日

查看全文
  • Abdelaziz Sahbani,
  • Kamel Ben Saad

摘要

The appropriate design of an energy management system (EMS) is crucial in fuel cell hybrid electric vehicles (FCHEVs) as it must ensure an efficient coordination of the multiple energy sources including the fuel cell (FC), battery, and super-capacitor (SC). This paper presents an up-to-date and extensive review of EMS strategies, focusing on RL-based methodologies and their application on new powertrain architectures. RL is identified as a promising technology that could be used to implement intelligent and adaptive control solutions that would also be energy efficient for FCHEVs. The review focuses on four main aspects: (1) powertrain configurations adopted for FCHEVs, (2) reinforcement learning techniques, parameters, and performance measures, (3) the use of multi-objective optimization for EMS design, and (4) datasets and simulation cycles applied in the studies being reviewed. Special focus is placed on the ability of the RL framework to learn dealing with conflicting objectives. These insights aim to guide researchers in advancing next-generation EMS for FCHEVs.