锂离子电池剩余寿命间接预测方法  

Indirect remaining useful life prognostics for lithium-ion battery

作　　者：庞景月 马云彤[1] 刘大同[1] 彭宇[1] 

机构地区：[1]哈尔滨工业大学电气工程及自动化学院自动化测试与控制系,哈尔滨150080

出　　处：《中国科技论文》2014年第1期28-36,共9页China Sciencepaper

基　　金：国家自然科学基金资助项目(61301205);高等学校博士学科点专项科研基金资助项目(20112302120027);哈尔滨工业大学科研创新基金资助项目(HIT.NSRIF.2014017)

摘　　要：针对锂离子电池在线剩余寿命预测时容量难以直接测量以及预测表达的不确定性等问题,提出一种利用锂离子电池充放电监测参数构建剩余寿命预测健康因子的方法框架,实现了锂电池健康状态的表征,同时利用高斯过程回归(Gaussian process regression,GPR)方法给出剩余寿命预测的不确定性区间,从而构建了锂离子电池在线剩余寿命预测的方法体系。基于NASA锂离子电池数据集和卫星锂离子试验数据的剩余寿命预测验证和评估实验,表明本文提出的方法框架可以很好地支撑电池在线剩余寿命预测的应用,具备较好的电池剩余寿命预测精度和不确定性管理能力。The capacity is often used as the health index （HI）for predicting the remaining useful life （RUL）of a lithium-ion bat-tery.However,it is quite challenging to monitor and estimate the capacity of a lithium-ion battery on-line.Thus,a framework of RUL estimation for lithium-ion battery is proposed by extracting HI with charging and discharging monitoring parameters.As a result,the health state can be indicated with the extracted HI.Moreover,the Gaussian process regression （GPR）algorithm is u-tilized for RUL estimation,with which the confidence interval for the RUL value can be obtained,and then the prediction struc-ture of RUL for lithium-ion battery can be built.The experiments are implemented with the NASA battery data set and satellite testing data set.Experimental results prove that the proposed RUL prediction framework can realize the on-line application,and achieve satisfactory performance in prediction precision and prognostic uncertainty management for lithium-ion battery.

关 键 词：测试计量技术及仪器 锂离子电池 故障预测和健康管理 剩余寿命预测 高斯过程回归 健康因子 

分 类 号：TM912[电气工程—电力电子与电力传动]