基于证据推理规则CS-SVR模型的锂离子电池SOH估算  

State of Health Estimation of Lithium-ion Battery Using a CS-SVR Model Based on Evidence Reasoning Rule

作　　者：徐宏东 高海波[1] 徐晓滨[2] 林治国[1] 盛晨兴[1] XU Hongdong;GAO Haibo;XU Xiaobin;LIN Zhiguo;SHENG Chenxing(School of Energy and Power Engineering,Wuhan University of Technology,Wuhan 430063,China;Institute of System Science and Control Engineering,Hangzhou Dianzi University,Hangzhou 310018,China)

机构地区：[1]武汉理工大学能源与动力工程学院,武汉430063 [2]杭州电子科技大学自动化学院,杭州310018

出　　处：《上海交通大学学报》2022年第4期413-421,共9页Journal of Shanghai Jiaotong University

基　　金：国家自然科学基金项目(U1709215);国家水运安全工程技术研究中心开放基金(A2019003);新能源船舶设计研发及应用示范关键技术研究(19DZ1203100)。

摘　　要：锂离子电池健康状态(SOH)的准确性影响电池的安全性和使用寿命.针对锂离子电池SOH估算问题,提出一种基于证据推理(ER)规则的布谷鸟搜索支持向量回归(CS-SVR)的SOH估算模型,并利用NASA Ames研究中心的锂离子电池数据集进行SOH估算试验.该方法以电池放电循环的平均放电电压和平均放电温度为模型输入,利用ER规则进行推理,得到输入数据的融合信度矩阵.将该矩阵输入CS算法优化的SVR模型得到电池SOH估算结果.结果表明,与5种估算效果较好的现有模型相比,基于ER规则的CS-SVR模型具有更良好的估算性能.The state of health(SOH) estimation accuracy of lithium-ion battery affects the safety and service life of batteries. Aimed at the problem in SOH estimation of lithium-ion battery, a cuckoo search support vector regression(CS-SVR) model based on the evidence reasoning(ER) rule was proposed for SOH estimation. The lithium-ion battery data from NASA Ames Center was used to perform the SOH estimation test. In this method, the average voltage and average temperature of battery discharge cycles were taken as model input, and a fusion belief degree matrix of input data was obtained by the ER rule. The SOH estimation result of the battery was obtained by inputting a fusion belief degree matrix into the SVR model optimized by the CS algorithm. The results show that the CS-SVR algorithm based on the ER rule has a better estimation performance than the five existing models.

关 键 词：锂离子电池 证据推理规则 布谷鸟搜索 健康状态估算 支持向量回归 

分 类 号：TM912.9[电气工程—电力电子与电力传动]