基于XGBoost算法的锂离子电池健康状态估算SOH Estimation of Li-ion Battery Based on XGBoost Algorithm
费陈,赵亮,王云恪,王树泉
FEI Chen,ZHAO Liang,WANG Yunke,WANG Shuquan
摘要(Abstract):
由于锂离子电池内部电化学反应的复杂性以及外部工作环境的不确定性,难以对电池SOH(健康状态)进行准确估算。为了提高SOH的估算精度,提出一种基于XGBoost算法的锂离子电池健康状态估算方法。首先,利用Matplotlib对锂离子电池数据进行分析,并提取平均电压、电压差、电流差和温度差等特征量来描述电池的老化过程。然后,利用XGBoost算法建立估算模型,对电池SOH进行智能估算。实验结果表明:与线性回归、随机森林、支持向量机、K近邻算法这4种回归算法相比,所提方法的估算精度和泛化能力更优,并且可将估算误差控制在±0.4%左右。
Due to the extremely complex electrochemical reactions inside the lithium-ion battery and the uncertainty of the external working environment,it is difficult to achieve an accurate estimation of SOH(state of health). To improve the accuracy of SOH estimation,an estimation method of lithium-ion battery health state based on XGBoost algorithm is proposed. First,Matplotlib is used to analyze the lithium-ion battery data and to extract the characteristic quantities such as average voltage,voltage difference,current difference and temperature difference to describe the aging process of the battery. Then,the XGBoost algorithm is used to build an estimation model for the intelligent SOH estimation of battery. The experimental results show that the proposed method has higher estimation accuracy and superior generalization ability compared with four regression algorithms,viz.,linear regression,random forest,support vector machine(SVM),and K-nearest neighbors algorithm,and the estimation error can be controlled within ±0.4%.
关键词(KeyWords):
锂离子电池;特征提取;健康状态估算;XGBoost算法
lithium-ion battery;feature extraction;SOH estimation;XGBoost algorithm
基金项目(Foundation):
作者(Author):
费陈,赵亮,王云恪,王树泉
FEI Chen,ZHAO Liang,WANG Yunke,WANG Shuquan
DOI: 10.19585/j.zjdl.202205003
参考文献(References):
- [1]王学远,李日康,魏学哲,等.基于传荷电阻的锂离子电池剩余寿命预测研究[J].机械工程学报,2021,57(14):105-117.
- [2]冯海林,张翾.基于新健康因子的锂电池健康状态估计和剩余寿命预测[J].南京大学学报(自然科学),2021,57(4):660-670.
- [3]潘斌,董栋,钱东培,等.磷酸铁锂电池内阻分量快速检测方法[J].浙江大学学报(工学版),2021,55(1):189-194.
- [4] ZHOU Q,WANG C,SUN Z Y,et al.Human-knowledgeaugmented Gaussian process regression for state-of-health prediction of lithium-ion batteries with charging curves[J].Journal of Electrochemical Energy Conversion and Storage,2021,18(3):15-19.
- [5] ZHANG Y J,LIU Y J,WANG J,et al.State-of-health estimation for lithium-ion batteries by combining modelbased incremental capacity analysis with support vector regression[J].Energy,2022(5):56-63.
- [6]梁培维,张彦会.基于欧姆内阻对锂电池健康状态的估算[J].电源技术,2019,43(10):1623-1625.
- [7]韦海燕,钟腾云,潘海鸿,等.基于改进HPPC锂离子电池内阻测试方法研究[J].电源技术,2019,43(8):1309-1315.
- [8]刘伟霞,田勋,肖家勇,等.基于混合模型及LSTM的锂电池SOH与剩余寿命预测[J].储能科学与技术,2021,10(2):689-694.
- [9]沈佳妮,贺益君,马紫峰.基于模型的锂离子电池SOC及SOH估计方法研究进展[J].化工学报,2018,69(1):309-316.
- [10]王萍,弓清瑞,张吉昂,等.一种基于数据驱动与经验模型组合的锂电池在线健康状态预测方法[J].电工技术学报,2021,36(24):5201-5212.
- [11]范文杰,徐广昊,于泊宁,等.基于电化学阻抗谱的锂离子电池内部温度在线估计方法研究[J].中国电机工程学报,2021,41(9):3283-3293.
- [12]宋明超,李国春,王丽梅,等.基于电化学阻抗谱的锂离子电池内部温度估算研究[J].农业装备与车辆工程,2020,58(5):56-61.
- [13]王倩倩.基于等效电路模型的车用动力电池内部状态估算研究[D].武汉:武汉理工大学,2018.
- [14]丁羿茗,吕瑞强,蒋超.基于分数阶锂电池模型的FOMIAEKF算法SOC估计[J].电源技术,2021,45(9):1148-1151.
- [15]李超然,肖飞,樊亚翔,等.基于卷积神经网络的锂离子电池SOH估算[J].电工技术学报,2020,35(19):4106-4119.
- [16]王义,刘欣,高德欣.基于BiLSTM神经网络的锂电池SOH估计与RUL预测[J].电子测量技术,2021,44(20):1-5.
- [17]何浩然,丁稳房,吴铁洲,等.基于IGA-BP神经网络的锂电池健康状态估算[J].电源技术,2022,46(1):78-82.
- [18]张天瑞,周福强,吴宝库,等.基于ARIMA和XGBoost的滚动轴承故障预测模型研究[J].制造技术与机床,2022(4):176-182.
- [19] ZHANG B,WEI Z Y,REN J D,et al.An empirical study on predicting blood pressure using classification and regression trees[J].IEEE Access,2018(6):21758-21768.
- [20]韦振汉.锂离子电池荷电及健康状态预测方法研究[D].桂林:广西师范大学,2018.
- [21]李靖,王志和,倪浩.基于改进下垂控制的直流微网运行研究[J].发电技术,2021,42(6):765-774.
- [22]温素芳,王生铁,田桂珍.基于混合储能系统的微电网并离网平滑切换控制策略研究[J].电力电容器与无功补偿,2021,42(5):204-213.
- [23]李沂洹,李慷,余渐.锂离子电池荷电状态与健康状态估计方法[J].发电技术,2021,42(5):537-546.
- [24]柴雯,吴明锋,杨姝.能源互联网背景下电力储能技术发展问题研究[J].山西电力,2021(2):36-39.
- [25]王晓辉,郑超,张思义,等.融合算法在蓄电池SOC估算中的研究综述[J].四川电力技术,2021,44(1):43-46.
- [26]刘俊华,张启超,李程,等.磷酸铁锂电池模组健康度快速评估方法研究[J].电网与清洁能源,2020,36(10):112-118.
- [27]于童.梯次利用电池在储能电站中应用的可行性分析[J].山西电力,2020(4):11-13.