基于自适应无迹卡尔曼滤波算法的锂离子动力电池状态估计

States Estimation of Li-ion Power Batteries Based on Adaptive Unscented Kalman Filters

应用传统的无迹卡尔曼滤波(unscented Kalman filter,UKF)算法估计电动汽车锂离子动力电池核电状态(state of charge,SOC)时,常会出现由于电池模型参数不准确而造成估计误差增大的问题,该文采用了自适应无迹卡尔曼滤波(adaptive unscented Kalman filter,AUKF)算法解决该问题。AUKF算法是一种循环迭代算法,可以实时估计电池模型中的欧姆内阻,提高电池模型准确性,从而提高电池SOC估计精度。另外,电池的欧姆内阻可以表征电池的健康状态(state of health,SOH),因此还可以根据电池的欧姆内阻估计出电池的SOH。在设定工况下对电池做充放电实验,实验分析表明,与UKF相比,AUKF提高了电池SOC估计的精度,并能准确的估计出电池的欧姆内阻。

When using the traditional unscented Kalman filter （UKF） to estimate the electric vehicle li-ion power battery state of charge （SOC）, the inaccurate battery model often causes estimation error to increase. Adaptive unscented Kalman filter （AUKF） was used to solve this problem in this paper. AUKF is a kind of cyclic iterative algorithm, and using it can estimate the inner ohmic resistance of the battery model in real time. Therefore, it improves the accuracy of the battery model, and thus further improves the accuracy of battery SOC estimation. In addition, the battery state of health （SOH） also can be estimated because the inner ohmic resistance of the battery can characterize the battery SOH. The battery charged and discharged experiments were done under setting conditions and the experimental analysis showes that AUKF improves the estimation accuracy of battery SOC compared with UKF, and AUKF can accurately estimate the inner ohmic resistance of the battery.