基于机器学习的锂电池寿命预测方法  

Machine learning based approach for predicting the lifespan of lithium batteries

作　　者：鲁译徽 王玲芝 LU Yihui;WANG Lingzhi(School of Automation,Xi'an University of Posts and Telecommunications,Xi'an 710121,China)

机构地区：[1]西安邮电大学自动化学院,西安710121

出　　处：《黑龙江电力》2024年第6期471-477,共7页Heilongjiang Electric Power

基　　金：国家自然科学基金面上项目(项目编号:52177194)。

摘　　要：为提高锂离子电池剩余使用寿命(RUL)预测的准确性,提出一种基于机器学习的锂电池寿命预测方法。选用美国国家航空航天局(NASA)预测中心的锂电池数据,将电池容量、放电至截止电压时间和温度作为模型的输入数据,根据数据的非线性特征和小样本规模的特点,选用支持向量回归(SVR)模型作为预测锂电池剩余使用寿命的核心模型。分别采用麻雀搜索算法(SSA)和蜣螂优化算法(DBO)对模型参数进行优化,建立SSA-SVR模型和DBO-SVR模型进行锂电池剩余寿命预测。通过2种优化后的模型与SVR模型的预测效果对比,验证了SSA-SVR模型和DBO-SVR模型的准确性与优越性,证明优化过后的模型比优化前具有更高的预测精度。同时,SSA-SVR模型的性能略优于DBO-SVR,因此更适合用于进行锂电池剩余寿命的预测。To enhance the prediction precision of remaining useful life(RUL)for lithium-ion batteries,a machine learning-based methodology for lithium battery lifespan forecasting was proposed.Data from the NASA Prediction Center regarding lithium batteries were selected,utilizing battery capacity,discharge time to cut-off voltage,and temperature as input parameters for the model.Based on the nonlinear characteristics and small sample size of the data,the support vector regression(SVR)model was selected as the core model for predicting the remaining useful life of lithium batteries.The sparrow search algorithm(SSA)and dung beetle optimization(DBO)algorithms were used to optimize the model parameters,and the SSA-SVR model and DBO-SVR model were established for remaining useful life prediction of lithium batteries.The accuracy and superiority of the SSA-SVR and the DPO-SVR models were validated through a comparison of their predictive performance with that of the SVR model.The optimized models demonstrate improved prediction accuracy comparing to their pre-optimized counterparts.Additionally,the SSA-SVR model exhibits slightly superior performance to the DPO-SVR model,making it more suitable for predicting the remaining life of lithium batteries.

关 键 词：锂电池 机器学习 支持向量回归模型 麻雀搜索算法 蜣螂优化算法 

分 类 号：TM911[电气工程—电力电子与电力传动]