基于多健康特征与TLSSA-SVM模型的锂电池SOH预测
SOH Prediction for Lithium Batteries Based on Multiple Health Features and TLSSA-SVM Model
司梦婷 1张梅 1徐云涛1
作者信息
- 1. 安徽理工大学 电气与信息工程学院,安徽 淮南 232001
- 折叠
摘要
为提高锂电池非线性退化过程中SOH的预测精度,提出了一种基于多特征融合与TLSSA-SVM模型的锂电池SOH预测方法.首先从电池原始数据中提取电池的多健康特征:直接测量特征、容量增量特征、相似性特征,通过多维尺度缩放,降低高维特征数据的信息重叠;接着引入Tent混沌映射和莱维飞行改进SSA,避免了SSA陷入局部最优;然后利用改进的TLSSA对SVM进行参数寻优,得到最优模型;最后利用该模型对锂电池进行SOH预测,并以NASA锂电池数据集为依据进行实验验证.结果表明,该预测方法的RMSE仅在1.52%以下,R2均在0.974 以上.
Abstract
In order to improve the prediction accuracy of SOH in the nonlinear degradation process of lithium batteries,this paper proposes a SOH prediction method for lithium batteries based on multi-feature fusion and TLSSA-SVM model.Firstly,the multi-health features of the battery are extracted from the original battery data:direct measurement features,capacity incremental features,and similarity features,and the information overlap of the high-dimensional feature data is reduced by multi-dimensional scale scaling;then Tent Chaos Mapping and Levy Flight are introduced to improve the SSA,which avoids the SSA from falling into the local optimum;and then parameter searching for the SVM is carried out by using the improved TLSSA to obtain the optimal model;finally,the model is used to predict SOH for lithium batteries,and the NASA lithium battery dataset is used as the basis for experimental verification.The experimental results show that the RMSE of the prediction method is just below 1.52%and the R2 are all above 0.974.
关键词
健康状态/动态时间规整距离/麻雀搜索算法/支持向量机/健康特征Key words
state of health/dynamic time warping distance/sparrow search algorithm/support vector ma-chines/health features引用本文复制引用
出版年
2024
NETL
NSTL
万方数据