基于分数阶模型的锂离子电池SOC估计

SOC estimation of Li-ion battery based on fractional order modeling

准确的荷电状态(SOC)估算,有助于延长电池寿命并确保电池安全。由于电荷转移阻抗和扩散阻抗对应的时间常数不同,电池模型参数也不同。研究基于分数阶模型自适应遗忘因子递推最小二乘法(FOM-AFFRLS)的参数辨识,以实时捕捉遗忘因子和参数的变化,并采用扩展卡尔曼滤波估计SOC。FOM-AFFRLS算法的误差为1%,小于分数阶基于遗忘因子的递推最小二乘法(FOM-FFRLS)、整数阶自适应遗忘因子递推最小二乘法(IOM-AFFRLS)和整数阶遗忘因子递推最小二乘法(IOM-FFRLS)等,验证所提方法在动态工况下正常工作时,具有较高的SOC估计精度。该方法能克服错误初始值引起的发散,SOC初值为0.7时,平均绝对误差小于0.068,鲁棒性较好。

Accurate state of charge(SOC)estimation helps to extend battery life and ensure battery safety.Due to the different time constants corresponding to charge transfer impedance and diffusion impedance,the battery model parameters are different.The fractional-order model-based adaptive forgetting factor recursive least squares(FOM-AFFRLS)method is investigated for parameter identification in order to capture the variation of forgetting factor and parameters in real time,and to estimate the SOC using extended Kalman filtering.The error of FOM-AFFRLS algorithm is 1%,which is smaller than that of fractional-order forgetting-factor-based recursive least squares(FOM-FFRLS),integer-order adaptive forgetting-factor recursive least squares(IOM-AFFRLS)and integer-order forgetting-factor recursive least squares(IOM-FFRLS).It verifies that the proposed method has high SOC estimation accuracy under dynamic operating conditions and normal operation.The method can overcome the dispersion caused by the wrong initial value,the average absolute error is less than 0.068 when the initial value of SOC is 0.7,the robustness is good.