基于ARX模型的锂离子电池荷电状态在线估算

Online State-of-Charge Estimation for Lithium-ion Batteries Based on the ARX Model

准确估计锂电池荷电状态（state-of-charge,So C）是电源管理系统的核心技术问题之一。针对锂离子电池等效电路模型参数难以获取这一关键问题,该文采用自回归各态历经（autoregressive exogenous,ARX）模型建立锂电池等效模型,由基于赤池信息量准则的遗传算法确定ARX模型的阶数,采用递推最小二乘法获取模型系数;然后利用得到的模型系数和锂电池状态方程构造自适应卡尔曼滤波（adaptive Kalman filter,AKF）算法所需方程,再由卡尔曼迭代方程求出锂电池SoC,文中将这种估计锂电池SoC的方法称为ARX-AKF算法。最后,通过多组对比实验,验证了该算法的有效性和准确性。实验结果表明：在混合动力脉冲能力特性实验和美国城市循环工况下,采用该算法的锂电池SoC估计误差分别在0.5%和0.8%以内,从而证实了该算法具有一定的工程应用价值。

The accurate state-of-charge （SoC） estimation of lithium batteries is a key technique in battery management systems. Aiming at solving the problem that the parameters of the lithium battery model are hard to be obtained, a new lithium-ion battery model is set up based on the autoregressive exogenous （ARX） model. The order and the coefficients of the ARX model are identified by the genetic algorithm with Akaike＇s information criterion （AIC） and the recursive least square method, respectively. And, the coefficients of the ARX model and the state equation of lithium batteries are transformed to the space state equation that needed in the Kalman filter algorithm, then the SoC can be obtained by the Kalman filter recurrence formula. The method of SoC estimation for lithium：ion batteries is named ARX-AKF method in this paper. Finally, the accuracy and reliability of the method are verified by the multiple comparison experiments. The experimental results show that under the condition of hybrid pulsepower characteristic （HPPC） and urban dynamometer driving schedule （UDDS）, the error of SoC estimation of the lithium-ion battery can be efficiently limited below 0.5% and 0.8%, respectively. So, the proposed ARX-AKF method has great potential in industrial application.