An Efficient Reconfigurable Battery Network Based on the Asynchronous Advantage Actor-Critic Paradigm

TY - JOUR

T1 - An Efficient Reconfigurable Battery Network Based on the Asynchronous Advantage Actor-Critic Paradigm

AU - Yang, Feng

AU - Meng, Jinhao

AU - Ci, Marvin

AU - Lin, Ni

AU - Gao, Fei

N1 - Publisher Copyright: © 2015 IEEE.

PY - 2025

Y1 - 2025

N2 - Although the battery systems continue to grow in electric vehicles (EVs), smart grids, and backup power systems due to their capability to supply power and energy, the current battery systems essentially are fixed which limits efficient energy usage. Recently, by employing the digital battery concept through energy digitization, traditional battery systems can be transformed into reconfigurable battery networks (RBNs). This RBN paradigm improves the inherent defects of the fixed battery system, which means the performance of the RBN will no longer be limited by the consistency of the cells. In this article, an adaptive control framework with the asynchronous advantage actor-critic (A3C) paradigm on performing online optimization for the dynamical RBN system is proposed. By utilizing its policy and asynchronous learning property, the proposed paradigm can improve the learning performance and the battery pack capacity through an improvement of cell consistency. It is noted that the proposed method is verified by both the simulation and experiment with data from an RBN.

AB - Although the battery systems continue to grow in electric vehicles (EVs), smart grids, and backup power systems due to their capability to supply power and energy, the current battery systems essentially are fixed which limits efficient energy usage. Recently, by employing the digital battery concept through energy digitization, traditional battery systems can be transformed into reconfigurable battery networks (RBNs). This RBN paradigm improves the inherent defects of the fixed battery system, which means the performance of the RBN will no longer be limited by the consistency of the cells. In this article, an adaptive control framework with the asynchronous advantage actor-critic (A3C) paradigm on performing online optimization for the dynamical RBN system is proposed. By utilizing its policy and asynchronous learning property, the proposed paradigm can improve the learning performance and the battery pack capacity through an improvement of cell consistency. It is noted that the proposed method is verified by both the simulation and experiment with data from an RBN.

KW - Asynchronous advantage actor-critic (A3C)

KW - large-scale

KW - model-free

KW - reconfigurable battery network (RBN)

UR - https://www.scopus.com/pages/publications/85194894669

U2 - 10.1109/TTE.2024.3406444

DO - 10.1109/TTE.2024.3406444

M3 - Article

AN - SCOPUS:85194894669

SN - 2332-7782

VL - 11

SP - 1479

EP - 1487

JO - IEEE Transactions on Transportation Electrification

JF - IEEE Transactions on Transportation Electrification

IS - 1

ER -