摘要
插电式柴电混合动力汽车具有多种工作模式,发动机频繁启停过程中会导致油耗增加和SCR催化器的效率降低,导致排放恶劣.以P2型插电式柴电混合动力汽车为研究对象,建立所需动力系统模型,将深度强化学习(Deep Reinforcement Learning,DRL)应用到插电式混合动力汽车能量管理中.采用TD3算法对PHEV油耗和排放进行综合优化,并将结果与动态规划算法(Dynamic Programming,DP)进行对比分析,结果表明:基于TD3算法的控制策略的油耗和NO_(X)排放量分别为2.477 L/km、0.2023 g/km,分别达到DP控制策略的94.1%和89.4%的控制效果,证明了提出的控制策略的有效性.
The plug in diesel electric hybrid vehicle has a variety of working modes.Frequent engine start and stop will increase fuel consumption and reduce the efficiency of SCR catalyst,resulting in poor emissions.In this paper,taking P2 plug-in diesel electric hybrid vehicle as the research object,the required powertrain model is established,and deep reinforcement learning(DRL)is applied to the energy management of plug-in hybrid vehicle.TD3 algorithm is used to comprehensively optimize the fuel consumption and emission of PHEV,and the results are compared with that of the dynamic programming(DP).The results show that the fuel consumption and NOX emission of the control strategy based on TD3 algorithm are 2.477 L/km and 0.2023 g/km respectively,and they achieves 94.1%and 89.4%of the control effect of DP control strategy respectively,which proves the effectiveness of the proposed control strategy.
作者
赵春领
吴化腾
ZHAO Chun-ling;WU Hua-teng(School of Mechatronics & Vehicle Engineering,Chongqing Jiaotong University,Chongqing 400074,China)
出处
《西安文理学院学报（自然科学版）》
2022年第1期25-34,共10页
Journal of Xi’an University(Natural Science Edition)
基金
重庆交通大学研究生科研创新项目(CYS20288)。
关键词
插电式混合动力汽车
深度强化学习
能量管理策略
TD3算法
plug in hybrid electric vehicle
deep reinforcement learning
energy management strategy
TD3 algorithm
