电池BATTERYBIMONTHLY第53卷图6车辆2诊断结果分析Fig.6Diagnosticresultanalysisofvehicle2从图6可知,22号电池电压由于没有触发目前BMS的欠压阈值报警和压差过大报警,未被检出,而所提方法可检测出故障电池并报警。4结论本文作者首先研究了VMD算法,将原始信号进行VMD分解,然后基于重构信号提取特征,最后利用LOF算法对特征进行诊断。经实车数据验证,所提方法能够在车辆热失控前21个采样点发出故障预警,且相较于BMS提前17个采样点。基于LLE的锂离子电池故障诊断有如下优势:在特征提取方面,无量纲特征参数在故障诊断领域表现出良好的稳定性和对故障信息的敏感性。利用LLE算法提取特征,既保留了二维空间的关键信息,消除了正常电池的不一致性影响,也降低了计算量,是仅通过传统无量纲参数无法实现的。参考文献:[1]朱景哲,张希,高一钊,等.数据驱动的锂离子电池智能故障诊断算法[J].电池,2022,52(4):401-405.ZHUJZ,ZHANGX,GAOYZ,etal.DatadrivenintelligentfaultdiagnosisalgorithmforLi-ionbattery[J].BatteryBimonthly,2022,52(4):401-405.[2]JIANGJC,LITY,CHANGC,etal.Faultdiagnosismethodforlithium-ionbatteriesinelectricvehiclesbasedonisolatedforestal-gorithm[J].JEnergyStorage,2022,50:104177.[3]ZHANGH,NIUGX,ZHANGB,etal.Cost-effectivelebesguesam-plinglongshort-termmemorynetworksforlithium-ionbatteriesdiagnosisandprognosis[J].IEEET-IE,2022,69(2):1958-1967.[4]李洪军,汪大春,杨哲昊,等.基于DCGAN的燃料电池故障诊断[J].电池,2022,52(5):502-506.LIHJ,WANGDC,YANGZH,etal.FuelcellfaultdiagnosisbasedonDCGAN[J].BatteryBimonthly,2022,52(5):502-506.[5]XIAB,SHANGYL,NGUYENT,etal.Acorrelationbasedfaultdetectionmethodforshortcircuitsinbatterypacks[J].JPowerSources,2017,337:1-10.[6]YAOL,FANGZP,XIAOYQ,etal.Anintelligentfaultdiagnosismethodforlithiumbatterysystemsbasedongridsearchsupportvectormachine[J].Energy,2021,214:118866.[7]YAOL,XIAOYQ,GONGXY,etal.Anovelintelligentmethodforfaultdiagnosisofelectricvehiclebatterysystembasedonwave-letneuralnetwork[J].JPowerSources,2020,453:227870.[8]YANGRX,XIONGR,HEHW,etal.Afractional-ordermodel-basedbatteryexternalshortcircuitfaultdiagnosisapproachforall-climateelectricvehiclesapplication[J].JCleanProd,2018,187:950-959.[9]柏云耀,邹时波,李...
