基于Apollo-BiGRU模型的磷酸铁锂电池SOC估算  

SOC estimation of lithium iron phosphate batteries based on Apollo-BiGRU model

作　　者：张荣达 任佳乐 刘素贞[2,3] ZHANG Rongda;REN Jiale;LIU Suzhen(State Power Investment Corporation Hebei Electric Power Co.,Ltd.,Shijiazhuang 050000,China;State Key Laboratory of Reliability and Intelligence of Electrical Equipment(Hebei University of Technology),Tianjin 300130,China;Hebei Key Laboratory of Electromagnetic Field and Electrical Reliability(Hebei University of Technology),Tianjin 300130,China)

机构地区：[1]国家电投集团河北电力有限公司,河北石家庄050000 [2]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 [3]河北省电磁场与电器可靠性重点实验室(河北工业大学),天津300130

出　　处：《电工电能新技术》2024年第11期22-31,共10页Advanced Technology of Electrical Engineering and Energy

基　　金：国家自然科学基金项目(51977058);河北省自然科学基金项目(E2024202010)。

摘　　要：由于磷酸铁锂电池开路电压与SOC之间曲线较为平坦,电信号对于SOC变化敏感度较低,严重影响估算精度。超声技术可以检测因材料物理性质改变而引起的电池声学特性差异,进而表征电池状态。本文提出一种基于高相关超声特征及Apollo优化算法的BiGRU网络模型的磷酸铁锂电池SOC估算方法。首先,开展电池充放电过程中超声检测实验,基于波形和统计角度提取并选择高相关性超声时频域特征;然后,对比多种先进数据驱动模型及优化算法,研究基于Apollo-BiGRU深度网络模型的SOC估算方法;最后,在不同电流倍率下实现该方法的准确性和可靠性验证。研究结果表明,不同电流倍率下SOC估算的平均绝对误差和均方根误差分别低于1.26%和1.46%,验证了该方法的可行性。Due to the relatively flat curve between the open-circuit voltage and SOC of LiFePO 4 battery,the electrical signal is less sensitive to the change of SOC,which seriously affects the estimation accuracy.Ultrasound technology can detect differences in the acoustic properties of batteries due to changes in the physical properties of the material and thus characterize the battery state.In this paper,a method for SOC estimation of lithium iron phosphate battery based on BiGRU network model with high correlation ultrasonic characteristics and Apollo optimization algorithm is proposed.First,ultrasonic detection experiments during battery charging and discharging were carried out,and highly correlated ultrasonic time-frequency domain features were extracted and selected based on waveform and statistical perspectives.Then,comparing a variety of state-of-the-art data-driven models and optimization algorithms,and the SOC estimation method based on the Apollo-BiGRU deep network model was studied.Finally,the validation of the accuracy and reliability of the method was realized at different current multiplicities.The results show that the average absolute error and root mean square error of SOC estimation at different current multiplicities are lower than 1.26%and 1.46%,respectively,which verifies the feasibility of the method.

关 键 词：磷酸铁锂电池 荷电状态估算 超声特征 Apollo-BiGRU 不同倍率 

分 类 号：TM912[电气工程—电力电子与电力传动]