基于充电过程的锂电池SOH估计和RUL预测

State of health estimation and remaining useful life predication of lithium batteries using charging process

车用锂离子电池的健康状态(state of health,SOH)和剩余寿命(remaining useful life,RUL)是锂离子电池的关键状态参数,为实现其准确的预估以保障整车安全可靠的运行,基于电动汽车充电过程提出一种改进高斯过程回归(Gaussian process regression,GPR)的锂电池SOH估计和RUL预测模型。首先以最大互信息系数(maximal information coefficient,MIC)、Pearson系数筛选充电过程的多元信息作为健康因子,基于主成分分析(principal components analysis,PCA)简化模型结构并使用粒子群算法和组合核函数改进高斯过程回归,实现车用锂离子电池SOH的准确在线估计以及RUL预测。通过NASA锂离子电池数据集验证了模型的有效性:测试电池SOH估计的最大均方根误差(root mean square error,RMSE)为0.0148,SOH预测的最大RMSE为0.0169,RUL预测的最大绝对误差为1个循环次数。

The state of health(SOH)and the remaining useful life(RUL)of lithium-ion batteries for vehicles are key state parameters.Based on the charging process of electric vehicles,an improved Gaussian process regression(GPR)model for lithium battery SOH estimation and RUL prediction is proposed to achieve accurate estimation of SOH and RUL of the battery to ensure the safe and reliable operation of the vehicle.First,the maximal information coefficient(MIC)and Pearson coefficient are used to screen the multivariate information of the charging process as health factors.Further,the model structure is simplified using principal components analysis(PCA).The Gaussian process regression is then improved using particle swarm optimization and the combined kernel function.Finally,accurate online estimation of SOH and prediction of future RUL and SOH are realized.The validity of the model is verified with the NASA lithium-ion battery data set.This model outperforms other studies in terms of estimation and prediction accuracy.The maximum root mean square error(RMSE)of SOH estimation for test batteries is 0.0148,the maximum RMSE of SOH prediction is 0.0169,and the maximum absolute error of RUL prediction is 1 cycle.