欢迎访问《汽车安全与节能学报》，
汽车安全与节能学报

汽车安全与节能学报 ›› 2025, Vol. 16 ›› Issue (1): 159-169.DOI: 10.3969/j.issn.1674-8484.2025.01.016

• 智能驾驶与智慧交通 • 上一篇    下一篇

网联混合动力汽车队列的生态驾驶与能量管理分层控制

张富椿1(), 尹燕莉1,*(), 马永娟2, 肖杭洋1, 陈海鑫1, 余凯1   

  1. 1.重庆交通大学 机电与车辆工程学院，重庆 400074，中国
    2.中汽研汽车检验中心（昆明）有限公司，昆明651700，中国
  • 收稿日期:2024-09-17 修回日期:2024-10-13 出版日期:2025-02-28 发布日期:2025-03-04
  • 通讯作者: * 尹燕莉，副教授。E-mail：cqu_ylyin@126.com。
  • 作者简介:张富椿（1999—），男（汉），重庆，硕士研究生。E-mail：1320431797@qq.com。
  • 基金资助:
    贵州大学现代制造技术教育部重点实验室开放项目(GZUAMT2023KF[9]);四川省新能源汽车智能控制与仿真测试技术工程研究中心资助项目(XNYQ2022-003);城市轨道交通车辆系统集成与控制重庆市重点实验室开放课题(CKLURTSIC-KFKT-212005);重庆交通大学研究生科研创新项目(2024S0082)

Ecological driving and hierarchical control of energy management for networked hybrid electric vehicle queues

ZHANG Fuchun1(), YIN Yanli1,*(), MA Yongjuan2, XIAO Hangyang1, CHEN Haixin1, YU Kai1   

  1. 1. School of Mechatronics and Vehicle Engineering, Chongqing Jiaotong University, Chongqing 400074, China
    2. China Automotive Research and Development Center (Kunming) Co., Ltd, Kunming 651700, China
  • Received:2024-09-17 Revised:2024-10-13 Online:2025-02-28 Published:2025-03-04

RichHTML

3

PDF

56

摘要：

为解决智能网联环境下混合动力汽车（HEV）队列通过连续交通信号灯路口时的舒适和经济性问题，提出一种基于网联HEV队列的生态驾驶与能量管理分层控制方法。上层控制器针对连续交通信号灯路口建立目标车速规划模型；根据目标车速范围建立纵向约束限制，建立以安全、舒适、跟随、经济和通过性为指标的目标函数；并采用模型预测控制（MPC）算法求解多目标函数获得最优车速。下层控制器采用深度强化学习（DQN）算法优化混合动力汽车能量管理，将上层求解的最优车速作为下层输入获取发动机电机的最优输出。结果表明：该文所提控制策略可以保证汽车队列的行驶安全，生态驾驶汽车队列平均油耗比普通队列降低了8.51%，在避免停车等待的同时改善了乘坐舒适性和燃油经济性。

关键词: 混合动力汽车（HEV）, 汽车队列, 生态驾驶, 交通信号灯, 模型预测控制（MPC）, 深度强化学习

Abstract:

In order to solve the problem of comfort and economy of hybrid electric vehicle (HEV) queues passing through continuous traffic signal intersections in a smart grid environment, a hierarchical control method of ecological driving and energy management based on networked HEV queue was proposed. The upper-level controller developed a target speed planning model for intersections with continuous traffic lights. Based on the defined target speed range, longitudinal constraint limits were established, and an objective function encompassing safety, comfort, following behavior, economy, and passability was formulated. The multi-objective function was solved using a model predictive control (MPC) algorithm to determine the optimal vehicle speed. Meanwhile, the lower controller adopted deep reinforcement learning (DQN) algorithm to optimize the energy management of the HEV, and took the optimal speed solved by the upper layer as the input of the lower layer to obtain the optimal output of the engine motor. The results show that the proposed control strategy can ensure the driving safety of the car queue, and the average fuel consumption of the eco-driving car queue is reduced by 8.51% compared with that of the ordinary queue, improving the ride comfort and fuel economy while avoiding the parking wait.

Key words: hybrid electric vehicle (HEV), vehicle queue, eco-driving, traffic signal, model predictive control, deep reinforcement learning (DQN)

中图分类号: