【摘要】 锂离子电池的荷电状态估计(SOC)是电池管理系统(BMS)的关键指标，准确的SOC预测是锂电池安全工作的关键保证.针对由电池模型的参数固定而导致模型参数辨识准确性不够以及传统无迹卡尔曼滤波精度较低、稳定性差等问题，运用自适应遗忘因子递推最小二乘算法(AFFRLS)对二阶RC等效电路模型进行在线参数辨识，结合自适应无迹卡尔曼滤波算法(AUKF)联合估计电池荷电状态.实验结果表明，AFFRLS-AUKF联合算法能够自适应多个工况下的SOC估计，在DST工况下SOC的平均误差降低至0.003 5;在FUDS工况下SOC的平均误差降低至0.011 0、在US06工况下SOC的平均误差降低至0.001 1、在BJDS工况下SOC的平均误差降低至0.007 7.该算法解决了在多个工况下锂电池因参数时变而导致的估计精度较低的问题，为锂离子电池的使用寿命和管理系统的运行效率提供了保障.更多还原

【Abstract】 The state of charge(SOC) estimation of lithium-ion batteries was a key indicator of the battery management system(BMS), and accurate SOC prediction was a crucial guarantee for the safe operation of lithium batteries.This paper addressed issues such as insufficient accuracy in model parameter identification due to fixed battery model parameters, as well as the low precision and poor stability of traditional unscented Kalman filtering. An adaptive forgetting factor recursive least squares(AFFRLS) algorithm was employed for online parameter identification of a second-order RC equivalent circuit model, combined with an adaptive unscented Kalman filter(AUKF) algorithm for joint SOC estimation.Experimental results showed that the AFFRLS-AUKF joint algorithm could adaptively estimate SOC under multiple working conditions. Under the DST condition, the average SOC error was reduced to 0.003 5; under the FUDS condition, it was reduced to 0.011 0; under the US06 condition, it was reduced to 0.001 1; and under the BJDS condition, it was reduced to 0.007 7.This algorithm resolved the problem of low estimation accuracy caused by time-varying battery parameters under multiple working conditions, providing a guarantee for the service life of lithium-ion batteries and the operational efficiency of the management system.更多还原