基于弛豫电压模型的锂离子电池RUL预测

RUL prediction of Li-ion battery based on relaxation voltage model

锂离子电池在长期使用过程中呈非线性退化趋势,预测非线性退化对延长电池寿命和确保安全意义重大。提出一种利用弛豫电压作为特征序列的非线性退化拐点预测方法,进行拐点和剩余使用寿命(RUL)的联合预测,建立结合拐点退化特征的RUL预测框架,提高预测精度。通过迁移学习,在不同的电池数据集上验证所提联合预测方法的性能,拐点和RUL预测的平均绝对误差在26次循环内,均方根误差低于28次循环。该方法利用弛豫电压来预测拐点和RUL,从而间接预测电池健康状态(SOH),具有预测精度好、应用范围广等特点。

Li-ion battery exhibits a nonlinear degradation trend over long-term use.Predicting nonlinear degradation is crucial for extend battery life and ensure safety.A nonlinear degradation knee-point prediction method using relaxation voltage as a feature sequence is proposed,which enables joint prediction of knee-point and remaining useful life(RUL).A framework for RUL prediction combining knee-point degradation features is established to improve prediction accuracy.The proposed joint prediction method is validated on different battery datasets using transfer learning,the mean absolute error for knee-point and RUL prediction is below 26 cycles,the root mean squared error is below 28 cycles.This method uses relaxation voltage to predict knee-point and RUL,thereby indirectly predicting the state of health(SOH)of battery,with advantages in prediction accuracy and broad applicability.