强化学习
适应性
燃料效率
能源消耗
能源管理
化石燃料
计算机科学
汽车工程
动态规划
工程类
模拟
能量(信号处理)
人工智能
算法
电气工程
生态学
统计
数学
生物
废物管理
作者
Weiqi Chen,Jue Peng,Tong Ren,Hailong Zhang,Haibo He,Chunye Ma
标识
DOI:10.1016/j.enconman.2023.117685
摘要
Ecological driving (eco-driving) is a promising technology for transportation sector to save energy and reduce emission, which works by improving vehicle behaviors in traffic scenarios. Fuel cell hybrid electric vehicles (FCHEV) are receiving extensive attentions due to global fossil energy crisis, but whose implementations for eco-driving result in multiple objective collaborative optimization problems. In this paper, an eco-driving framework for FCHEV is proposed based on deep deterministic policy gradient (DDPG) algorithm. And it combines adaptive cruise control (ACC) and energy management strategy (EMS) into an integrated architecture. Firstly, in order to achieve excellent balance between driving behaviors and fuel economy, an appropriate weight coefficient value is determined after adequate explorations. Secondly, power-varying equivalent hydrogen conversion coefficient function is constructed to save fuel consumption by 8.97%. Thirdly, ablation experiments for health state of fuel cell system present 19.95% decrease in terms of health degradation. Then, comparison experiments indicate that the DDPG-based eco-driving strategy can reach 94.16% of that of dynamic programming with respect to equivalent hydrogen consumption, meanwhile with best ride comfortability. Moreover, simulation results under validation driving cycle manifest its excellent adaptability.
科研通智能强力驱动
Strongly Powered by AbleSci AI