广西科技大学学报2025，Vol.36Issue(4)：67-73,7.DOI:10.16375/j.cnki.cn45-1395/t.2025.04.009
基于短时充电片段的实车动力电池SOH估计
SOH estimation of real vehicle power batteries based on short-time charging segments
摘要
Abstract
Driven by major strategic decisions such as carbon peaking and carbon neutrality,the production and sales scale of new energy vehicles have continuously reached new historical highs.To address the issue of accurately estimating the state of health(SOH)of the battery during real vehicle operation,the current maximum available capacity of the battery was calculated using the ampere-hour integration method based on short-term charging segment data of real vehicles,and box plots were used to eliminate capacity outliers caused by factors such as sensor noise.Battery degradation features were extracted based on easily accessible data fields such as current,voltage,temperature,and the state of charge(SOC),and the correlation between each feature and the health status was analyzed using correlation coefficient analysis.Principal component analysis was used to reduce the dimensionality of feature parameters to lower the computational complexity.A real vehicle power battery health status prediction model was constructed based on the long short-term memory neural network,and the optimal model hyperparameters were determined using the grey wolf optimizer(GWO).The results show that based on the monitoring data in the 80%-90%SOC range during real vehicle charging,the absolute error of the model in predicting the battery health status is 0.27 A·h,and the model goodness of fit is 0.89,which can accurately estimate the health status of real vehicle power batteries.关键词
实车数据/动力电池/健康状态(SOH)/相关性分析/长短期记忆神经网络/灰狼优化算法(GWO)Key words
real vehicle data/power battery/state of health(SOH)/correlation analysis/long short-term memory neural network/grey wolf optimizer(GWO)分类
信息技术与安全科学引用本文复制引用
陈嘉铭,唐文俊,何山,黄鹏,张占喜..基于短时充电片段的实车动力电池SOH估计[J].广西科技大学学报,2025,36(4):67-73,7.基金项目
规模化电动汽车与电网互动关键技术研究与示范应用(二期)项目(090000KK52222138)资助 （二期）
