基于在线参数辨识和ASR-UKF的锂离子电容SOC估计

State of charge co-estimation of lithium-ion capacitor based on online parameter identification and ASR-UKF

以一种新型混合型超级电容器——锂离子电容为研究对象,针对其在混合动力机车应用中的SOC估计问题,建立锂离子电容的二阶等效电路模型,采用带遗忘因子的递推最小二乘法(FFRLS)和自适应平方根无迹卡尔曼滤波算法(ASR-UKF)交叉联合的方法对锂离子超级电容的荷电状态(SOC)进行估算。FFRLS可以对动态变化的模型参数进行实时且精确的在线辨识,在获得精确的模型参数的基础上,运用ASR-UKF算法对SOC估计不断修正更新,消除系统的未知噪声所引起的误差,并且利用协方差平方根来代替协方差矩阵进行迭代运算,克服滤波发散的问题,进而获得最优的SOC估算值。通过在实验室环境下的混合脉冲功率特性(HPPC)工况和模拟工况的实验仿真,评估了该联合算法的有效性。

A new type of supercapacitor,lithium-ion capacitor(LIC),was used as the research object.Aiming at the state of charge(SOC)estimation problem of LIC in hybrid electric vehicle application,a second-order equivalent circuit model of lithium-ion capacitors was established.The forgetting factor recursive least-squares method(FFRLS)and adaptive square root unscented Kalman filter algorithm(ASR-UKF)co-estimation algorithm was used to estimate the LIC SOC.FFRLS could perform real-time and accurate online identification of dynamically changing model parameters.Based on the accurate model parameters,the ASR-UKF was used to update SOC,eliminating errors caused by unknown noise in the system.The square root of the covariance instead of the covariance matrix was used to perform the iterative operation to overcome the problem of filtering divergence,obtaining the optimal estimation value of SOC.The effectiveness of the proposed method was evaluated through experiments under hybrid pulse power characteristic conditions and simulation conditions in the laboratory environment.