Capacity Estimation of Serial Lithium-ion Battery Pack Using Dynamic Time Warping Algorithm

Abstract

The existence of the consistency degradation of the battery pack hinders the accurate estimation of pack capacity and cell capacity in the battery pack. The paper focuses on the capacity estimation of cells in the serial battery pack. The shape invariance of the charging voltage curve is discussed and used as the theoretical foundation of cell capacity difference identification. The matching relationship between two voltage curves is obtained based on the dynamic time warping algorithm. Then the capacity difference identification algorithm to calculate the capacity difference between the two cells is proposed. Based on the algorithm, a three-step capacity estimation method is established. The proposed method can only use the previous charging curve of one cell in the pack and the current charging data of the battery pack to rapidly estimate the capacity of each cell in the battery pack. A 16 serial LiFePO4 battery pack is employed to verify the method. The result shows the estimation error of cell capacities is less than 3% rated capacity. With this method, the cell capacities in the pack can be rapidly and accurately estimated, providing a foundation for the consistency analysis and equalization of the battery pack.