A novel energy management strategy for hybrid electric vehicles using deep reinforcement incentive learning

Ozan Yazar, Serdar Coskun, Fengqi Zhang, Lin Li, Cong Huang, Peng Mei and Hamid Reza Karimi

Energy, 2025, vol. 334, issue C

Abstract: Hybrid electric vehicles (HEVs) represent a significant step toward achieving fuel-efficient and environmentally friendly mobility. An effective energy management strategy (EMS) is needed to maximize the performance and energy efficiency of HEVs. However, since existing EMSs are generally developed utilizing rule-based or fixed structures, they often suffer from problems such as insufficient generalization capacity under variable driving conditions and exhibiting limited real-time applicability. Integrating deep reinforcement learning (DRL) algorithms into EMS design has attracted attention to overcome these limitations in recent years. DRL-based methods have a strong potential to develop more efficient and adaptive control strategies in HEVs. However, the proper design of reward functions representing multi-objective control goals and prioritizing the instantaneous objective remains a significant technical challenge, which can negatively affect the optimization performance during the learning process. We propose a deep reinforcement incentive learning (DRIL) algorithm to minimize fuel consumption and maintain battery charge sustainability under various driving conditions. To ignite the reinforcement learning agent to efficiently learn the optimal power distribution, we first propose an incentive reward function based on fuel consumption minimization and battery charge sustainability. Next, we introduce a novel soft actor-critic algorithm combined with incentive-based reward, forming the DRIL algorithm, which optimally balances exploration and exploitation for power allocation. Finally, the generalized advantage of the DRIL algorithm is evaluated through a human-in-the-loop (HIL) test-driving cycle, demonstrating its real-time applicability. The results of the experiments demonstrate that the proposed incentive mechanism can reduce fuel costs by 3.47%–3.04% compared to the existing DRL algorithms under the pre-training driving cycle and the HIL-obtained driving cycles, respectively.

Keywords: Hybrid electric vehicles; Energy management; Optimal power allocation; Incentive learning; Deep reinforcement learning (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S0360544225032360
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:334:y:2025:i:c:s0360544225032360

DOI: 10.1016/j.energy.2025.137594

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 ().