基于AR-ECM平均差异模型的串联电池组SOC、容量多尺度联合估计方法  

Multi-scale Joint Estimation of SOC and Capacity of Series Battery Pack Based on AR-ECM Mean-difference Model

作　　者：刘芳[1] 余丹 苏卫星 卜凡涛 LIU Fang;YU Dan;SU Weixing;BU Fantao(Tianjin Key Laboratory of Autonomous Intelligence Technology and Systems(Tiangong University),Xiqing District,Tianjin 300387,China;Neusoft Reach Automotive Technology,Co.,Ltd.,Shenyang 110000,Liaoning Province,China)

机构地区：[1]天津市自主智能技术与系统重点实验室(天津工业大学),天津市西青区300387 [2]东软睿驰汽车技术(沈阳)有限公司,辽宁省沈阳市110000

出　　处：《中国电机工程学报》2024年第10期3937-3948,I0016,共13页Proceedings of the CSEE

基　　金：国家重点研发计划项目(2021YFB2501800)。

摘　　要：考虑电池单体老化差异所致的电池组不一致性,针对串联电池组荷电状态(state of charge,SOC)、容量估计问题,提出一种基于自回归等效电路模型(autoregression equivalent circuit model,AR-ECM)的平均差异模型(mean-difference model,MDM)。基于此模型,提出串联电池组SOC、容量多尺度联合估计算法。该算法由2个部分组成,一是基于AR-ECM的MDM及差异化模型参数辨识策略:条件辨识策略和定频分组辨识策略;二是基于多时间尺度H无穷滤波(multi-timescale H infinity filter,Mts-HIF)的电池组SOC、容量联合估计算法。通过将所提出MDM中的自回归平均模型(autoregression mean model,AR-MM)与传统MDM中的n阶RC平均模型(nRC mean model,nRC-MM)比较,结果表明所提出的AR-MM在复杂运行工况下具有更优的动态跟随性能。依据最小化信息量准则(akaike information criterion,AIC),AR-MM具有更优的复杂度与精度的权衡。通过与基于多时间尺度扩展卡尔曼滤波(multi-timescale extended Kalman filter,Mts-EKF)联合状态估计算法比较,结果表明所提出的Mts-HIF状态估计算法具有更优的鲁棒性、精度和收敛速度。Considering the inconsistency of battery pack caused by aging difference of battery cell,and aiming at the state of charge(SOC)and capacity estimation issues of series battery pack,a mean-difference model(MDM)based on autoregressive equivalent circuit model(AR-ECM)is proposed.Based on this model,a multi-scale joint estimation algorithm of SOC and capacity of series battery pack is put forward.The algorithm includes two parts:one is the proposed MDM based on AR-ECM and the differentiated parameter identification strategy,including conditional identification strategy and fixed frequency grouping identification strategy.The other is the proposed joint estimation algorithm of battery pack SOC and capacity based on multi-timescale H infinity filter(Mts-HIF).In the experiment part,the autoregressive mean model(AR-MM)in the proposed MDM is compared with the nRC mean model(nRC-MM)in the traditional MDM,and the results show that the proposed AR-MM has better dynamic following performance under complex operating conditions.According to the Akaike information criterion(AIC),AR-MM has a better tradeoff between complexity and accuracy.Comparing with the joint state estimation algorithm based on multi-timescale extended Kalman filter(Mts-EKF),the Mts-HIF-based state estimation algorithm has better robustness,accuracy and convergence speed.

关 键 词：串联电池组 自回归等效电路模型 平均差异模型 容量 荷电状态 H无穷滤波 

分 类 号：TM912[电气工程—电力电子与电力传动]