改进安时法结合神经网络估算锂离子电池SOC  

Estimating SOC of Li-ion battery by improved AH combined with neural network

作　　者：吴海东[1] 任晓明[1] 那伟 黄超[1] 

机构地区：[1]上海电机学院电气学院,上海201306 [2]上海航天电源技术有限责任公司,上海201615

出　　处：《电池》2016年第1期16-19,共4页Battery Bimonthly

基　　金：上海市自然科学基金(12ZR1411700);上海市教委优青项目(ZZSDJ12003);上海电机学院研究生创新项目(A1-0225-15-005-04)

摘　　要：采用BP神经网络对库仑效率进行训练并预测,将预测得到的库仑效率代入改进安时(AH)算法,再基于Moto Hawk进行设计,应用于地铁应急牵引电池组管理系统。以952国产A车为试验对象,结合实际运行情况对荷电状态(SOC)进行估算和分析。试验结果表明,所采用的方法比传统AH法估算精度误差提高4.9%。The Coulombic efficiency was taken into the improved Ah computation,then the improved AH computation was designed which was based on Moto Hawk and used in the subway of emergency traction battery management system. The 952 domestic A subway was taken as experimental target,state-of-charge（ SOC） could be analyzed and estimated combined with the actual operating conditions. The test results showed that the estimation accuracy had been increased 4. 9% due to the method used.

关 键 词：荷电状态(SOC) 电池组管理系统 改进安时(AH)算法 BP神经网络 

分 类 号：TM912[电气工程—电力电子与电力传动]