Scholarlink

<p>A rational energy management strategy (EMS) is pivotal to the optimal performance and operation of fuel cell hybrid vehicles (FCHV). In this study, a novel algorithm based on improved deep Q-network (DQN) is proposed for the EMS of FCHV. This EMS uses fuel cell (FC) as the main energy source, and takes into account the hydrogen consumption of FC, the degree of smooth operation of lithium-ion (Li-ion) battery, the degree of power aging, as well as introduces for the first time the following effect of fuel cell system (FCS) on the demanded power. In order to improve the performance of the DQN algorithm, Double DQN (DDQN) and Dueling DQN mechanisms are introduced in this study to form the Dueling and Double Deep Q-network (D3QN) algorithm. The D3QN algorithm effectively reduces the overestimation problem by separating the action selection and evaluation processes, and improves the learning stability and convergence speed. Simulation results show that the average hydrogen consumption of the proposed EMS is reduced by 11.1% compared with the DQN algorithm. The aging of FC is retarded by 27.9% under FTP-75 operating condition. It also greatly improves the effectiveness of FCS in following the vehicle demand power with an average enhancement of 40.0%. The novel energy management framework proposed in this paper will provide more ideas for future research, which is conducive to reducing the hydrogen consumption of fuel cell vehicles and extending their service lifetime.</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

Novel Energy Management Strategy for Fuel Cell Hybrid Vehicles Considering Power Following Based on Improved Deep Q-Network

  • 期刊论文

  • 发表日期：

    2025年7月9日

查看全文
  • Kuanyu Chang,
  • Dongfang Chen,
  • Zhiqiang Chen,
  • Taoheng Yin,
  • Song Hu,
  • Xiaoming Xu,
  • Pucheng Pei

摘要

A rational energy management strategy (EMS) is pivotal to the optimal performance and operation of fuel cell hybrid vehicles (FCHV). In this study, a novel algorithm based on improved deep Q-network (DQN) is proposed for the EMS of FCHV. This EMS uses fuel cell (FC) as the main energy source, and takes into account the hydrogen consumption of FC, the degree of smooth operation of lithium-ion (Li-ion) battery, the degree of power aging, as well as introduces for the first time the following effect of fuel cell system (FCS) on the demanded power. In order to improve the performance of the DQN algorithm, Double DQN (DDQN) and Dueling DQN mechanisms are introduced in this study to form the Dueling and Double Deep Q-network (D3QN) algorithm. The D3QN algorithm effectively reduces the overestimation problem by separating the action selection and evaluation processes, and improves the learning stability and convergence speed. Simulation results show that the average hydrogen consumption of the proposed EMS is reduced by 11.1% compared with the DQN algorithm. The aging of FC is retarded by 27.9% under FTP-75 operating condition. It also greatly improves the effectiveness of FCS in following the vehicle demand power with an average enhancement of 40.0%. The novel energy management framework proposed in this paper will provide more ideas for future research, which is conducive to reducing the hydrogen consumption of fuel cell vehicles and extending their service lifetime.