基于集成ELM算法的汽车锂离子电池充电状态评估  

Evaluation and Analysis of Charging State of Automotive Lithium Battery Based on Double-layer ELM Model

作　　者：秦龙 Qin Long(Henan Polytechnic College,Automotive College,Zhengzhou Henan 450015,China)

机构地区：[1]河南交通职业技术学院汽车学院,河南郑州450015

出　　处：《山西电子技术》2025年第1期53-55,共3页Shanxi Electronic Technology

基　　金：河南省科技攻关项目(232102240049);中国交通教育研究会2022-2024年度教育科学研究课题(JT2022YB316);河南省教育科学规划课题(2023YB0413)。

摘　　要：锂离子电池充电状态直接影响到电动汽车领域的经济效益,采用双层集成极限学习机(ELM)模型评估锂离子电池充电荷电状态(SOC)状态,通过分析电池健康特征实现充电状态的迭代计算。研究结果表明:计入老化SOC参考值表现出来明显的增加变化,当NASA介于3.9~4.0 V时,SOC和SOH之间达到了最大相关度,表现出了优异映射性能。比较LSTM和单独ELM算法,集成ELM预测的SOC预测均方根误差是最低的,证明了本文算法的准确性。该研究有助于提高锂离子电池的寿命,具有很高的使用价值。The charging state of lithium-ion batteries directly affects the economic benefits in the field of electric vehicles.The dual-layer integrated extreme learning machine(ELM)model is used to evaluate the charging state of lithium-ion batteries,and the iterative calculation of charging state is realized by analyzing the battery health characteristics.The results show that the reference value of the aging SOC shows a significant increase,and when the NASA is between 3.9 and 4.0V,the correlation between SOC and SOH reaches the maximum,showing excellent mapping performance.Compared with LSTM and ELM algorithm alone,the root-mean-square error of SOC prediction with integrated ELM prediction is the lowest,which proves the accuracy of the proposed algorithm.The research helps to improve the life of lithium-ion batteries and has high use value.

关 键 词：锂离子电池 充电状态 健康特征 集成极限学习机 

分 类 号：TM912.9[电气工程—电力电子与电力传动] TP301.6[自动化与计算机技术—计算机系统结构]