汽车工程学报2026，Vol.16Issue(3)：431-445,15.DOI:10.3969/j.issn.2095-1469.2026.03.09
基于强化学习的燃料电池重卡多时间尺度能量管理策略
Reinforcement Learning-Based Multi-Time-Scale Energy Management Strategy for Fuel Cell Heavy-Duty Trucks
摘要
Abstract
To improve the fuel economy and fuel cell lifespan of fuel cell hybrid heavy-duty trucks,this paper proposes a multi-time-scale energy management strategy that comprehensively accounts for equivalent hydrogen consumption,fuel cell degradation,and power battery aging,while introducing a multi-time-scale adaptation mechanism.Though this time-scale decoupling design,the strategy effectively balances the transient response flexibility of the system with the overall economic objective.First,a reinforcement learning strategy is trained under the NEDC driving cycle to systematically investigate the impact of fuel cell power maintenance on the performance of the hierarchical strategy.Finally,simulation validation is performed under the WLTC testing cycle.The results show that,compared with the rule-based,ECMS,and short-time-scale DDPG strategies,the proposed strategy reduces equivalent hydrogen consumption by 4.5％,0.6％,and 3.5％,respectively.Furthermore,it effectively mitigates the degradation rates of both the fuel cell and the lithium-ion battery,thereby reducing the overall operating cost of the fuel cell hybrid truck.关键词
燃料电池重卡/多时间尺度/深度强化学习/能量管理Key words
fuel cell heavy-duty truck/multi-timescale/deep reinforcement learning/energy management分类
交通工程引用本文复制引用
刘福建,陈梁,祝乔,董大伟..基于强化学习的燃料电池重卡多时间尺度能量管理策略[J].汽车工程学报,2026,16(3):431-445,15.基金项目
重庆市技术创新与应用发展专项(cstc2019jscx-fxydX0020):金属双极板燃料电池模块电堆研发及附件系统匹配设计应用 （cstc2019jscx-fxydX0020）
