• CN:11-2187/TH
  • ISSN:0577-6686

机械工程学报 ›› 2017, Vol. 53 ›› Issue (16): 86-93.doi: 10.3901/JME.2017.16.086

• 特邀专栏:汽车先进动力系统的设计、优化与控制(下) • 上一篇    下一篇

基于进化-增强学习方法的插电式混合动力公交车能量管理策略

陈征1, 刘亚辉2, 杨芳1   

  1. 1. 宁波工程学院理学院 宁波 315211;
    2. 清华大学汽车安全与节能国家重点实验室 北京 100084
  • 收稿日期:2016-09-18 修回日期:2017-01-20 发布日期:2017-08-20
  • 通讯作者: 刘亚辉(通信作者),男,1980年出生,博士。主要研究方向为人-车系统动力学及混合动力汽车。E-mail:liuyahui@tsinghua.edu.cn E-mail:liuyahui@tsinghua.edu.cn
  • 作者简介:陈征,男,1977年出生,博士。主要研究方向为混合动力车辆能量管理策略。E-mail:jeffchenzheng@yahoo.com;杨芳,女,1978年出生,博士,副教授。主要研究方向为非线性系统控制。E-mail:liusha_02@163.com
  • 基金资助:
    国家自然科学基金资助项目(61503205,51375009)

Energy Management Strategy for Plug-in Hybrid Electric Bus with Evolutionary-Reinforcement Learning Method

CHEN Zheng1, LIU Yahui2, YANG Fang1   

  1. 1. Science School, Ningbo University of Technology, Ningbo 315211;
    2. State Key Laboratory of Automotive Safety and Energy, Tsinghua University, Beijing 100084
  • Received:2016-09-18 Revised:2017-01-20 Published:2017-08-20

摘要: 插电式混合动力客车越来越多出现在城市公交领域。为了更好地提升车辆的燃油经济性,整车能量管理策略成为一大研究热点。提出一种基于进化-增强学习方法的插电式混合动力公交车能量优化管理策略。首先,给出简化的车辆模型并基于增强学习系统给出能耗的优化目标函数;其次,针对此优化目标函数给出了初始的控制策略种群,并用进化算法求出最优的能量控制策略和最优能耗值;最后,通过仿真分析验证了算法的有效性。提出的新方法相对传统的电量消耗-维持(Charge depleting charge sustaining,CDCS)策略减少了大约12%的花费。

关键词: 插电式混合动力公交车, 进化算法, 能量管理策略, 同轴并联, 增强学习

Abstract: A growing number of plug-in hybrid electric buses(PHEB) emerge in the city traffic field. In order to achieve better economical performance of PHEB, the study of energy management strategy(EMS) for PHEB is become the focus of the research. This paper proposes a method of EMS for PHEB based on reinforcement learning. First of all, a simplified vehicle model is presented, and the function of optimal energy consumption is obtained through the reinforcement learning system. Secondly, initial population of control strategy respect to the cost function, and optimal control strategy and energy consumption value are obtained using evolutional algorithm. Finally, simulation results are presented to illustrate the effectiveness of this method. The cost of the energy consumption via novel strategy in this paper decrease about 12% compared with traditional CDCS strategy.

Key words: energy management strategy, evolutionary algorithm, plug-in hybrid electric bus, reinforcement learning, single-shaft parallel

中图分类号: