Abstract: Lithium-ion batteries are the main source of power for new energy vehicles, and their status of health estimation is closely related to battery maintenance, repair, and replacement. In this paper, battery capacity is selected as a direct health factor, three indirect health factors are extracted based on battery charging voltage curve, and the correlation between them and capacity is calculated using Pearson and Spearman correlation coefficients. The results show that the correlation between charge time equal voltage and capacity is the highest. Gaussian process regression (GPR) algorithm is selected to establish the state of health (SOH) estimation model, and conjugate gradient method (CGM) is selected to calculate the hyper parameter of the model, which can better estimate the SOH of lithium-ion batteries.

Key words: Lithium-ion batteries for vehicles; SOH; Health index; Gaussian process regression

摘要： 锂离子电池作为新能源汽车的主要动力来源，其健康状态估计和电池的保养、维护与 更换息息相关。文章选择电池容量作为直接健康因子，基于电池充电电压曲线提取 3 个间接 健康因子，采用 Pearson 和 Spearman 相关系数计算其与容量之间的相关性。结果发现，等电 压上升充电时间与容量的相关性最大，选取高斯过程回归（GPR）算法建立电池健康状态 （SOH）估计模型，选用共轭梯度法（CGM）计算模型超参数，能够对锂离子电池 SOH 进行 较好的估计。

关键词: 车用锂电池；SOH；健康因子；高斯过程回归