基于滑动窗口机制的贝叶斯优化算法在锂电池模型等效参数辨识中的应用

打开文本图片集

中图分类号：TP273；TM912 文献标识码：A 文章编号：2096-4706（2025）23-0126-08

Abstract：To improve the parameter identificationaccuracyof the second-orderRCequivalentcircuit model for lithiumionbateriesandsolvethe problems thattraditional methodsaresusceptibleto interferenceanddificult tobalanceaccuracy and efficiencyundercomplex working conditions，this paper proposes a parameter identification method based onan improved Bayesianoptimizationalgorithm.Basedon thesecond-orderRCmodel，baterycharacteristicdata with SOCrangingfrom 100% （204号 to 10% is generatedthrough HPPC working conditiontests.The coreofthe algorithm integrates a sliding window mechanism，a hybridsamplingstrategy，ndanadaptiveexitmechanism，andrealizesparameteridentificatiobycombiningwiththeByesian optimization framework，whichiscompared with multiple algorithms.Experimentalresultsshow that theimproved Bayesian optimizationalgorithmachievesasignificantimprovementcomparedwiththetraditionalBayesianoptimizationalgorithm.Under HPPC working conditions， the voltage RMSE is reduced by approximately 60% ， the overall maximum voltage error is reduced by more than 70% ，and the error oscilation amplitude at the moment of current mutation peaks is significantly smaller. The convergence efficiency is improved by approximately 60% ，which effectively enhances the calculation accuracy and efficiency ofthetraditional Bayesian optimization algorithmand provides areliable modelbasis forthestateestimationand eficient management of lithium-ion batteries.

Keywords：lithium-ionbatery;parameterdentification;second-orderRCmodel; Bayesianoptimizationalgorithm;HPPC working condition

0 引言

在全球能源结构向可持续转型的浪潮中，新能源产业已成为推动经济社会发展的核心动力之一，而锂离子电池作为能量存储与转换的关键载体，凭借其高能量密度、长循环寿命、低自放电率及无记忆效应等显著特性，已深度渗透至电动汽车、大规模储能系统、便携式电子设备等诸多领域[1-2]。（剩余13272字）