Multi-Agent Deep Reinforcement Learning-Based Multi-Objective Cooperative Control Strategy for Hybrid Electric Vehicles

2024/03/08 by Jiongpeng Gan, Shen Li, Xianke Lin +1 · 1 citation
Engineering · #Electric Vehicles and Infrastructure #Electric and Hybrid Vehicle Technologies #Transportation and Mobility Innovations

paper · doi:10.1109/tvt.2024.3373906

Abstract

The progress of artificial intelligence (AI) technology promotes the development of energy management research for hybrid electric vehicles (HEV). To explore the application of multi-agent deep reinforcement learning (DRL) algorithm in multi-objective cooperative control of HEVs, this article proposes a multi-objective cooperative control strategy based on the multi-agent deep deterministic policy gradient (MADDPG) algorithm for a four-wheel drive (4WD) HEV equipped with a hybrid energy storage system (HESS). Firstly, the parameters of the HESS of the 4WD HEV are matched. Secondly, the DRL-based regenerative braking control strategy and HESS power distribution strategy are designed. Thirdly, an HEV control method based on MADDPG is proposed, in which different agents are used for training different control strategies for cooperative control. The results show that the MADDPG-based multi-objective cooperative control strategy can achieve a better cooperative optimization effect than the single-agent DDPG-based multi-objective cooperative control strategy.

Cited by

• Machine learning-based approach for reduction of energy consumption in hybrid energy storage electric vehicle. [europepmc]

Related

• Energy Management of Multi-mode Hybrid Electric Vehicles based on Hand-shaking Multi-agent Learning
• Energy Management of Multi-mode Plug-in Hybrid Electric Vehicle using Multi-agent Deep Reinforcement Learning
• Human-like Energy Management Based on Deep Reinforcement Learning and Historical Driving Experiences
• Recent Progress in Energy Management of Connected Hybrid Electric Vehicles Using Reinforcement Learning
• Transfer Deep Reinforcement Learning-enabled Energy Management Strategy for Hybrid Tracked Vehicle