基于电池组模型和聚类算法的锂离子电池组SOC不一致估计

Estimation of SOC Inconsistencies in Lithium-ion Battery Packs Based on Battery Pack Modeling and Clustering Algorithm

限于单体的功率和能量,电池组必须由成百上千的单体串并联而成,同时依赖于有效的电池管理技术,从而保证车辆行驶的动力性和经济性。然而制造误差和使用环境的不同导致电池单体存在不一致性,这使得准确估计所有单体电池荷电状态(State of charge,SOC)变得困难,从而导致电池发生过充和过放。因此,融合电池组模型和聚类算法,提出一种新的锂离子电池组SOC不一致估计方法,在保证精度的同时极大地降低计算复杂度。基于充电数据特征利用二分k-means算法将众多单体分为不同等级,减小需要考虑的对象;建立考虑参数不一致的电池组模型,用于估计单体SOC;从仿真数据和实车数据两方面验证估计结果的精度和计算复杂度,并与现有模型对比,结果表明所提方法单体SOC估计的误差在0.03以内,计算效率提高3~6倍。

Limited by the power and energy of a single cell,a battery pack must consist of hundreds or even thousands of cells in series and parallel,and at the same time rely on an effective battery management system to ensure enough driving power and economy of vehicles.However,due to the manufacturing process and operating environment,there exist parameter inconsistencies among battery cells,which brings challenges to accurately estimating the state of charge(SOC)of all the cells in the pack,and results in over-charge and over-discharge.To estimate the SOC inconsistencies in lithium-ion battery packs,a new method is proposed by combing the battery pack modeling and the clustering algorithm,which can greatly reduce the computational complexity while ensuring accuracy.Based on the features of charging data,the bisecting k-means algorithm is adopted to categorize cells into different levels,reducing the objects that need to be considered.Then,a battery pack model considering parameter discrepancies is established to estimate SOC values of cells.The estimation accuracy and computational complexity have been verified by using both simulation and real-test data,and compared with existing models.The results indicate that the SOC estimation error of a single cell is less than 0.03 by using the proposed method,and the computational efficiency can be improved by 3-6 times.