Real-time Energy Optimization of Hybrid Electric Vehicle in Connected Environment Based on Deep Reinforcement Learning. Issue 10 (2021)

• Record Type: Journal Article
• Authors: He, Weiliang; Huang, Ying
• Is Part Of: IFAC-PapersOnLine. Volume 54:Issue 10(2021)
• Page Start: 176
• Page End: 181
• Publication Date: 2021
• Subjects: Powertrain control -- Connected -- automated vehicles -- Hybrid electric vehicles -- Vehicle-to-everything -- Deep deterministic policy gradient; Automatic control -- Periodicals; 629.805
• DOI: 10.1016/j.ifacol.2021.10.160
• Languages: English
• ISSNs: 2405-8963

Abstract

Abstract: In this paper, a real-time control method of hybrid electric vehicle is proposed based on rule-based speed planning and deep deterministic policy gradient (DDPG) energy management algorithm. This method can optimize fuel economy in real time based on all traffic information in a connected environment, and satisfy the constraints of driving safety and driving time. The results show that the proposed deep reinforcement learning algorithm DDPG can achieve lower fuel consumption. In addition, the proposed speed planning algorithm will not violate traffic rules and has good results.