An energy management strategy of deep reinforcement learning based on multi-agent architecture under self-generating conditions

Chengcheng Chang, Wanzhong Zhao, Chunyan Wang and Zhongkai Luan

Energy, 2023, vol. 283, issue C

Abstract: To improve the driving efficiency of hybrid power vehicle, an energy management strategy of deep reinforcement learning based on multi-agent architecture under self-generating vehicle driving conditions is proposed. Firstly, the kinematics segments are self-generated based on the Wasserstein generative adversarial network. The generator network G is used to generate kinematics segments. The discriminator network D is used to judge the credibility of the generated kinematics segments with the Wasserstein distance. The speed distribution characteristics of the training conditions and verification conditions established based on the self-generated segments are verified. Afterward, a multi-agent algorithm based on twin delayed deep deterministic policy gradient algorithm for hybrid systems is proposed by introducing centralized training with decentralized execution framework. The engine and a motor are used as two independent agents respectively. Different reward functions are designed based on training objectives to establish a mutually beneficial relationship of cooperation-restraint between the two agents. A driving mode constraint is designed in the environment to improve sample utilization. Finally, the simulation results demonstrate that our method can achieve better performance compared with other existing works.

Keywords: Hybrid electric vehicle; Energy management; Generative adversarial network; Multi-agent architecture; Deep reinforcement learning (search for similar items in EconPapers)
Date: 2023
References: View references in EconPapers View complete reference list from CitEc
Citations: View citations in EconPapers (1)

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S0360544223019308
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:energy:v:283:y:2023:i:c:s0360544223019308

DOI: 10.1016/j.energy.2023.128536

Access Statistics for this article

Energy is currently edited by Henrik Lund and Mark J. Kaiser

More articles in Energy from Elsevier
Bibliographic data for series maintained by Catherine Liu ().