SOC Estimation for Series-Connected Lithium-Ion Battery Pack Based on Clustering Algorithm

State of charge (SOC) estimation of power battery pack is critical for electric vehicles (EVs). During the service life of electric vehicles, the consistency of power pack changes due to manufacturing errors and different usage environments, which in turn affects the accuracy of battery pack SOC estimation. To improve the accuracy of pack SOC estimation while reducing the computational complexity, this paper combines clustering algorithm and mean-difference (M-D) model to propose a SOC estimation method considering the battery pack inconsistency. Based on the features of charging data, a hierarchical clustering algorithm is used to assemble the cells with similar SOC. After grouping, the M-D model considering the SOC inconsistency is applied to the SOC estimation of battery pack. It is verified by experiments that the proposed method ensures high accuracy while simplifying the computational complexity.