Soh Balancing Control Method for Drb Systems Based on Deep Reinforcement Learning

Abstract

The issue of battery balancing has always been a challenging problem in battery energy storage systems, especially when considering the State of Health (SOH) and State of Charge (SOC) of the battery comprehensively. It is preferable for retired batteries to balance their SOH since it can prolong the system lifetime and improve the overall deliverable capacity of the system. Currently, there is still no effective method that can address the SOH balancing issue in dynamical reconfigurable battery (DRB) systems adequately.This paper introduces a battery SOH balancing control method from a new perspective of a latent indicator named RSOC based on deep reinforcement learning (DRL), which considers the optimization goal of maximizing the deliverable capacity of the system while ensuring SOH balancing in the DRB system.Extensive case studies have been carried out in the latter part of the article, and the results show the effectiveness of the proposed control method.

Keywords: DRB, DRL, RSOC, SOC, SOH