A New Active Equalization Method for Series Battery Packs Based on Dynamic Clustering and Greedy Algorithm

To address the inconsistency issue in large-scale series-connected lithium battery packs, this paper proposes an active equalization topology along with a novel equalization control strategy. First, a modular active equalization structure based on a bidirectional flyback converter is introduced, which supports multiple operating modes and enables efficient cross-module energy transfer through a time-shared equalizer and a common equalization bus. To further maximize the performance of the equalization system, an innovative control strategy integrating dynamic clustering and a greedy algorithm is proposed. An improved DBSCAN clustering algorithm is employed to dynamically identify target battery groups, while the greedy algorithm is adopted to optimize the equalization path and improve the equalization speed. Experimental and Monte Carlo simulation results demonstrate that the proposed strategy respectively reduces the equalization time by 31.9%, 35.6% and 32.9% under static, charging and discharging conditions, compared with the conventional Mean-Difference algorithm. These findings verify that the proposed topology and control strategy not only mitigate battery inconsistency effectively but also significantly enhance the equalization speed.