基于LOF和改进Elman的锂电池健康状态估计  

ESTIMATION OF LITHIUM-ION BATTERY HEALTH STATE USING LOF AND ENHANCED ELMAN NETWORK

作　　者：曹铭[1] 曾骥 谢世坤 邱嵩 张文祥 付艳恕[1] CAO Ming;ZENG Ji;XIE Shikun;QIU Song;ZHANG Wenxiang;FU Yanshu(School of Advanced Manufacturing,Nanchang University,Nanchang,Jiangxi 330031,China;School of Mechanical and Electrical Engineering,Jinggangshan University,Ji’an,Jiangxi 343009,China;BYD Auto Industry Company Limited,Shenzhen,Guangdong 518118,China)

机构地区：[1]南昌大学先进制造学院,江西南昌330031 [2]井冈山大学机电工程学院,江西吉安343009 [3]比亚迪汽车工业有限公司,广东深圳518118

出　　处：《井冈山大学学报(自然科学版)》2025年第2期97-106,共10页Journal of Jinggangshan University (Natural Science)

基　　金：国家自然科学基金项目(51762034)。

摘　　要：实现锂离子电池健康状态的准确估计,对于提高电池组的安全性能和使用性能、延长电池寿命以及提高能源利用率是至关重要。为了提高电池健康状态估计的准确性和鲁棒性,提出基于LOF异常值检测和改进Elman网络的电池健康状态估计方法。通过容量增量分析提取IC曲线峰值、峰值位置、峰左右斜率四个特征,然后作为模型输入,利用LOF算法对输入数据异常值进行检测和处理,并采用SCA算法和集成学习Bagging方法,实验设计对Elman网络进行改进。最后,使用NASA数据集进行模型验证,结果表明,LOF-SCA-Elman-Bagging模型的估计平均均方根误差为1.04%,与BP、SVM、GPR相比,具有更高的估计准确性和鲁棒性。Accurate estimation of the health state of lithium-ion batteries is essential for enhancing the safety and usability of battery packs,ex-tending their lifespan,and improving energy utilization.To this end,a new method for estimating battery health,based on LOF outlier detection and an enhanced Elman network,is proposed.This method involves extracting four key features from the IC curve-peak,peak position,and slope around the peak-through capacity increment analysis.The LOF algorithm is employed to detect and process outliers in the input data,while the SCA algorithm and Bagging technique via integrated learning are used to refine the Elman network.Validation of the model using the NASA dataset shows that the LOF-SCA-Elman-Bagging model achieves an estimated average root-mean-square error of 1.04%,demonstrating higher accuracy and robustness compared to the traditional methods such as BP,SVM,and GPR.

关 键 词：电池健康状态 容量增量分析 神经网络 局部异常因子 鲁棒性 

分 类 号：TM912[电气工程—电力电子与电力传动]