文档名：基于观测方程重构滤波算法的锂离子电池荷电状态估计
摘要：滤波算法中观测方程的准确性在电池状态评估中起着决定性作用.然而,该文通过试验发现,由于温度、工作电流和荷电状态(SOC)的影响,即使使用精度较高的电池模型,扩展卡尔曼滤波(EKF)算法中观测方程的输出值与实际电压之间仍会存在较大误差,即产生了较大的新息.该文提出一种基于观测方程重组的增强型扩展卡尔曼滤波(E-EKF)算法.该算法的核心思想是利用具有温度、SOC和电流自适应能力的误差修正策略对观测方程进行重组,实现算法中新息的降低,进而提高SOC估计的准确性.使用两种不同温度下的典型工况试验对E-EKF算法的性能进行了验证.试验结果表明,该算法能够适应不同的温度和工况,并具有较高的SOC估计精度.
Abstract：Lithium-ionbatteriesplayacrucialroleinenergystorage,aerospace,newenergyvehiclesandotherfields.However,over-chargingandover-dischargingwillcauseirreversibledamage,consequently,toensurethereliableandsafeoperationofbatteries,itisverynecessarytoaccuratelyestimatethestateofcharge(SOC)inrealtime.ConsideringthattheobservationequationofthedynamicalsystemisanimportantfactoraffectingtheSOCestimationresult,anadaptiveerrorcorrectionstrategyisputforwardinthepaper,furthermore,theobservationequationisreconstructed.CombinedwiththeextendedKalmanfilter(EKF),theenhancedEKFisestablishedtoestimateSOC.Here,thewidelyusedfirst-orderRCmodelisadoptedasthebatteryequivalentcircuitmodel(ECM).Inaddition,itmainlystudiesthecommonlyused20%～80%SOCworkingrange.Firstly,theeffectsoftheparameteridentificationmethod,ambienttemperature,SOCandcurrentratioontheoffsetbetweentheoutputoftheobservationequationandthetruemeasurementvaluearestudied.Theresultsshowthat:(1)atthesametemperatureandworkingcondition,theprobabilitydensitydistributionoftheoutputerroroftheobservationequationproducedbythetwoparameteridentificationmethods(Recursiveleastsquaresofforgettingfactor(FFRLS)andEKF)isverysimilar,morespecifically,thedifferenceoftheaverageoutputerrorofthemislessthan15%.(2)Temperaturehasagreatinfluenceontheoutputerrorofobservationequation.Generally,thelowerthetemperature,thegreatertheoutputerror.Thehigherthetemperature,thecloserandsmallertheerrorsunderthesameSOC.(3)Ingeneral,thelargerthecurrentrate,thelargertheoutputerroroftheobservationequation.Theextremepointofoutputerrorbasicallyoccurswhenthecurrentratechanges.(4)TheoutputerrorofobservationequationfluctuateswithSOC,anditsdistributionhasnoobviousrule.Baseontheexperimentalanalysis,itisinferredthat,theoutputerroroftheobservationequationismainlyaffectedbythebatteryrate,temperatureandSOC,andislessaffectedbytheparameteridentificationalgorithm,soitwillbeignoredinthepaper.Secondly,accordingtotheoutputerrordistributionandanalysisresults,anadaptiveerrorcorrectionmodelisgenerated,whichisasecond-orderpolynomialaboutSOC.Inwhich,thecoefficientsofsecondaryandprimarytermsareaffectedbytemperatureandSOC,whilethecoefficientoftheconstanttermareaffectedbytemperatureandcurrentrate.Asaresult,theerrorcorrectionmodelcanadapttocurrentrate,SOCandtemperatureinpracticalapplications.Finally,theadaptiveerrorcorrectionmodelisusedtoreconstructtheobservationequation,combinedwiththeEKF(namedE-EKF)toupdatetheSOCinrealtime.DSTandUS06conditionsareusedtoverifythevalidityoftheadaptiveerrorcorrectionmodelandtheperformanceofE-EKFalgorithm.Inwhich,DSTconditionsat0℃,20℃and40℃areusedtotraintheadaptiveerrorcorrectionmodel,andUS06andDSTat0℃,10℃,20℃,25℃,30℃and40℃areusedtotestthemodel.Ontheonehand,theaccuracyoftheobservationequationofE-EKFandEKFisobtained,theresultsshowthat,comparedwithEKF,theaverageoutputerrorgeneratedbytheobservedequationinE-EKFisreducedby71.84%inDSTconditionand60.92%inUS06condition.Ontheotherhand,theSOCestimationaccuracyofE-EKF,EKFandadaptiveEKF(AEKF)isobtained.Theresultsshowthat,comparedwithEKFandAEKF,theaverageSOCestimationerrorofE-EKFisreducedby18.12%and6.15%inDSTconditionand73.64%and30.00%inUS06conditionrespectively.Insummary,currentrate,SOCworkingrangeandambienttemperaturewhichinfluencetheperformanceofLithium-ionbatteriesaredynamicchangeinpracticalapplications.Ignoringthesefactors,whenusingEKFtoestimateSOC,therewillbealargeerror(thatis,alargeinnovation)betweentheoutputoftheobservationequationandthemeasurementvalue,resultinginlowerSOCestimationaccuracy.Theadaptiveerrorcorrectionmodelproposedcanhelptoreducetheinnovation,soastoimprovetheaccuracyofSOCestimation.
作者：黄凯 孙恺 郭永芳 王子鹏 李森茂 Author：HuangKai SunKai GuoYongfang WangZipeng LiSenmao
作者单位：省部共建电工装备可靠性与智能化国家重点实验室(河北工业大学)天津300130河北工业大学人工智能与数据科学学院天津300130
刊名：电工技术学报
Journal：TransactionsofChinaElectrotechnicalSociety
年，卷(期)：2024, 39(7)
分类号：TM912
关键词：扩展卡尔曼滤波算法 误差修正方程 观测方程重组 SOC估计
Keywords：ExtendedKalmanfilteralgorithm(EKF) errorcorrectionstrategy reconstructionofobservationequation SOCestimation
机标分类号：TM912U469.72V448.25+3
在线出版日期：2024年4月12日
基金项目：河北省自然科学基金面上资助项目基于观测方程重构滤波算法的锂离子电池荷电状态估计[
期刊论文] 电工技术学报--2024, 39(7)黄凯 孙恺 郭永芳 王子鹏 李森茂滤波算法中观测方程的准确性在电池状态评估中起着决定性作用.然而,该文通过试验发现,由于温度、工作电流和荷电状态(SOC)的影响,即使使用精度较高的电池模型,扩展卡尔曼滤波(EKF)算法中观测方程的输出值与实际电压之间仍...参考文献和引证文献
参考文献
引证文献
本文读者也读过
相似文献
相关博文
基于观测方程重构滤波算法的锂离子电池荷电状态估计 State of Charge Estimation of Lithium-Ion Battery Based on Enhanced Extended Kalman Filter Algorithm with Observation Equation Reconstruction
基于观测方程重构滤波算法的锂离子电池荷电状态估计.pdf
- 文件大小:
- 5.36 MB
- 下载次数:
- 60
-
高速下载
|
