Sharif University of Technology Researchers Develop Multi-Agent Reinforcement Learning Approach for Continuous Battery Cell-Level Balancing

Strategic Insights -

Researchers from Sharif University of Technology have proposed a multi-agent reinforcement learning (MARL) approach for continuous cell-level balancing in lithium-ion battery packs, particularly for electric vehicles (EVs). This approach enables decentralized, cooperative management and goal-oriented decision-making, overcoming the limitations of centralized and rule-based methods. The intelligent MARL approach manages the flow of charge between cells, utilizing a shared low-voltage bus and individual DC-DC converters to maximize pack capacity and minimize state-of-charge (SoC) variance.

Key Takeaways:

  • The MARL approach uses a multi-agent framework to treat each cell as an independent agent, enabling decentralized, cooperative management and goal-oriented decision-making.
  • The approach utilizes a shared low-voltage bus and individual DC-DC converters to manage the flow of charge between cells, maximizing pack capacity and minimizing SoC variance.
  • The MARL model is trained with the trust region policy optimization (TRPO) algorithm, ensuring stability and partial observability.
  • Simulations demonstrate the MARL model's superiority over baselines, showing a 28% improvement in battery performance and balancing speed up to three times faster under EV load profiles.
  • Yasaman Tavakol-Moghaddam and Mehrdad Boroushaki are the researchers involved in this study, with Tavakol-Moghaddam being the contact person for additional information.
  • The study concludes that this approach significantly enhances battery efficiency, contributing to longer-lasting, sustainable solutions through intelligent decentralized management.

Statistics:

  • 28% improvement in battery performance achieved through the MARL approach
  • 3 times faster balancing speed under EV load profiles
  • 104898: the article ID for the research in Results in Engineering
  • June 9, 2025: the date the news was published
  • Tehran, Iran: the location of the Sharif University of Technology
  • Electric Vehicles (EVs): the target application for the MARL approach
  • Multi-Agent Reinforcement Learning (MARL): the approach used for continuous battery cell-level balancing

Sources:

  • "A multi-agent reinforcement learning approach for continuous battery cell-level balancing." Results in Engineering, 2025,26():104898. (Results in Engineering - https://www.journals.elsevier.com/results-in-engineering)
  • Yasaman Tavakol-Moghaddam, Department of Energy Engineering, Sharif University of Technology, P.O. Box 14565-114, Tehran, Iran. Contact for additional information.