基于ICEEMDAN和改进Informer的锂电池SOH估计

打开文本图片集

DOI：10.16652/j.issn.1004-373x.2025.18.007 引用格式：，，，等.基于ICEEMDAN和改进Informer的锂电池SOH估计[J].现代电子技术，2025，48（18））：39-44.

关键词：锂电池；SOH预测；ICEEMDAN算法；Informer网络；容量回升；空洞因果卷积中图分类号：TN919-34；TM912 文献标识码：A 文章编号：1004-373X（2025）18-0039-06

Lithium battery SOH estimation based on ICEEMDAN and improved Informer

GUO Xinchun， ZHANG Yi， ZHANG Zhichao， LI Chunling （HebeiWind-Solar-Hydrogen-StorageSafety Monitoringand IntelligentOperation Centerof TechnologyInnovation （NorthChinaUniversityofScienceandTechnology），TangshanO6321O，China）

Abstract：Accuratelyestimating thestateofhealth （SOH）oflithium-ionbatteries （LIBs）cansignificantlyimprove thesafety of baterymanagement systemsandreduceoperationand maintenance（O&M）costs.Inorder toovercometheproblemsof predictioerrorscausedbythecapacityrecoveryandinsuffcientinformationcapturedbytraditionalmodels，aSOHestimation methodbasedontheICEEMDAN-DC-Informer modelisproposed.TheLIBscapacitydataaredecomposed intosub-modalities withdiffrentfrequenciesbymeansoftheimprovedcompleteensembleempiricalmodedecompositionwithadaptivenoise （ICEEMDAN）toeliminatetheerorcausedbythephenomenonofcapacityreboundinSOHestimation.TheInformermodelis improved byreplacingon-dimensionalconvolution withemptycausalconvolution，whichcanincrease theamountofcaptured informationwhilereducingthecomputationaloverhead.TheproposedmethodiscomparedwithInformer，ICEEMDAN-Informer andDCInformeronLIBsdatasetprovidedbyNASA.Incomparisonwiththecomparativealgorithm，theproposedmodelcan reduceMAE，RMSE，MAPE，andtimecostby35.6%，11%，26.4%，and15%，respectively.Theresultsindicatethat decomposing capacitywith ICEEMDANcaneliminatetheinterferenceofcapacityreboundphenomenon，and theproposed method has higher accuracy and lower time overhead.

Keywords：lithium batery；SOHprediction； ICEEMDANalgorithm；Informernetwork；capacityrecovery；null causal convolution

0 引言

锂离子电池具有高输出电压、长寿命和优异的能量密度等特点[1-3]，因此在储能领域被广泛应用。（剩余6793字）