Abstract: With the extensive development of large-scale energy storage systems, fast and accurate estimation of lithium-ion batteries' state of charge (SOC) is critical for safe and reliable system operation. However, under the traditional fixed series-parallel battery cell/module topology, the open-circuit voltage (OCV) of a battery cell/module cannot be directly measured. Therefore, it is unlikely to establish an OCV-SOC mapping relationship for accurate SOC estimation. This paper proposed a new method for accurate SOC estimation based on a novel dynamic reconfigurable battery (DRB) network. This DRB-based SOC estimation method could directly measure the OCV in seconds, and then the accurate SOC of a battery cell/module could be estimated by using the gradient boosting decision tree (GBDT). The experimental results show the proposed method's effectiveness and efficiency, which provides a paradigm-shifting framework for battery status estimation.