排
网络拓扑
控制器(灌溉)
工程类
扭矩
控制工程
控制理论(社会学)
电动汽车
能源管理
拓扑(电路)
计算机科学
能量(信号处理)
控制(管理)
功率(物理)
计算机网络
人工智能
电气工程
数学
统计
物理
量子力学
农学
生物
热力学
作者
Yanli Yin,Xuejiang Huang,Sen Zhan,Huan Gou,Xinxin Zhang,Fuzhen Wang
标识
DOI:10.1016/j.jclepro.2023.137414
摘要
To address the problem of longitudinal control and energy management of nonlinear hybrid electric vehicle platoon with different unidirectional communication topologies, a novel hierarchical energy management control strategy based on different communication topologies for platoon is proposed. Firstly, the upper platoon controller is based on the established nonlinear platoon longitudinal dynamics model. The vehicle-to-vehicle communication is applied to obtain the preceding vehicle information under different unidirectional communication topologies. The distributed model predictive control is used to realize the platoon longitudinal control and obtain the desired demand torque. Secondly, the lower energy management controller is based on desired demand torque and combines with current battery state of charge, the equivalent factor-based Deep Q-Learning is adopted to reasonably allocate the power component torque. Then, the platoon control effects of different unidirectional communication topologies are compared and verified under the given driving cycle. The preceding and leading following topology is chosen that satisfies real-time and optimality. The topology is also used to verify the effectiveness and working condition adaptability for this strategy. Finally, the simulation results under Chongqing actual working condition show that the strategy can well meet the requirements of platoon following, safety and comfort. The average fuel economy is improved by 2.61% and 7.58% compared with traditional deep Q-learning and Q-Learning, respectively, which can achieve global optimal similar to dynamic programming.
科研通智能强力驱动
Strongly Powered by AbleSci AI