基于可变阶模型和MIAUKF算法的锂电池SOC估计

SOC Estimation of Lithium Batteries Based on Variable-Order Models and the MIAUKF Algorithm

摘要: 新能源汽车的性能与安全，取决于锂电池荷电状态估算。针对传统等效电路模型固定阶数无法适配电池动态特性、传统滤波算法对非高斯噪声鲁棒性不足等问题，本文提出一种基于可变阶模型与多新息自适应无迹卡尔曼滤波(MIAUKF, Multi-input adaptive unscented Kalman filter)算法的联合估计方法。实验表明端电压最大误差仅10 mV，较传统二阶模型降低74%。在恒流放电工况下，SOC (state-of-charge)估计平均误差低至0.63%，最大误差较UKF和AUKF算法降低66%。本研究借助对模型、算法的改进，展现出了较高的精度，还具备强鲁棒性与实用性。

Abstract: The performance and safety of new energy vehicles depend on the state-of-charge estimation of lithium batteries. Aiming at the problems that the fixed order of the traditional equivalent circuit model cannot adapt to the dynamic characteristics of the battery, and the traditional filtering algorithm is not robust to non-Gaussian noise, this paper proposes a joint estimation method based on the variable-order model and the multi-innovation adaptive untraceable Kalman filter algorithm. Experiments show that the maximum error of terminal voltage is only 10 mV, which is 74% lower than the traditional second-order model. Under the constant current discharge condition, the average error of SOC estimation is as low as 1.61%, and the maximum error is 66% lower than that of UKF and AUKF algorithms, respectively. With the improvement of the model and algorithm, this study shows high accuracy and strong robustness and practicality.