基于自适应无迹卡尔曼滤波器的锂电池SOC估计

摘要

Estimating the state of charge (SOC) of lithium battery by traditional unscented Kalman filter (UKF) can obtain accurate results,however precondition of applying the UKF is that the statistical properties of the process noises and observation noises should be obtained accurately.An algorithm of adaptive UKF(AUKF) combined UKF and the adaptive filter is proposed.Firstly,a second-order equivalent circuit model of lithium battery is established.Secondly,one lithium iron phosphate battery is chosen as the testing object,and the parameters identification of the model is accomplished applying the least square method by some experimental data.And then,the detail steps of the AUKF algorithm based on unscented transformation(UT) are illustrated.The experimental results indicate that the AUKF can estimate SOC with error less than 1.2％ under the condition of both constant current and urban dynamometer driving schedule (UDDS) cycle.Compared with the traditional UKF,the proposed AUKF algorithm has a better accuracy and stronger tracking capabilities.%采用传统无迹卡尔曼滤波器(UKF)来估计锂电池荷电状态(SOC)的结果较为精确,但其应用前提是要精确获得系统过程噪声和观测噪声的统计特性.结合UKF与自适应滤波,提出一种自适应UKF(AUKF)算法,以二阶RC等效电路模型为基础,并以磷酸铁锂电池为测试对象,通过实验数据结合最小二乘法完成模型参数辨识,提出并详细给出基于无损交换(UT)自适应卡尔曼滤波器的算法步骤,测试实验结果表明:采用AUKF的算法估计锂电池SOC精度在恒流和美国城市循环工况(UDDS)动态工况下均能达到1.2％以内,相比传统的UKF算法具有更强的估计精度和自适应跟踪能力.