基于差分热伏安法的锂离子电池SOH多模型融合估计方法

Multi-model Fusion Estimation Method for Lithium-ion Battery SOH Based on Differential Thermal Voltammetry

针对电池老化分析中存在的问题,采用随机森林回归法进行多模型融合。首先,从差分热伏安曲线分析了电池老化过程的关键因素,提取5个健康因子作为一级输入。其次,利用支持向量回归和高斯过程回归模型分别对SOH进行初步预测。最后,利用随机森林模型进行初步融合。为了验证模型的有效性,在牛津电池退化数据集上进行了对比实验。与单模型估计方法相比,该方法在SOH估计中具有更好的精度和更强的鲁棒性,平均MAE和平均RMSE分别为0.46%和0.51%。

Aiming to address the challenges of battery aging analysis,this paper proposes a multi-model fusion method using random forest regression.Firstly,key factors in the battery aging process are analyzed based on differential thermal voltam⁃metry curves and five health factors are extracted as inputs.Second,support vector regression and Gaussian process regression models are used to predict the state of health(SOH).Finally,a random forest model is employed for preliminary fusion.To evaluate the proposed method,an experiment is conducted on the Oxford battery degradation dataset.Compared to the single model estimation approach,the proposed method achieves higher accuracy and robustness in SOH estimation,with mean MAE and RMSE of 0.46%and 0.51%,respectively.