基于多模型自适应卡尔曼滤波器的电动汽车电池荷电状态估计

Electric Vehicle Battery SOC Estimation Based on Multiple-model Adaptive Kalman Filter

基于电池的戴维宁(Thevenin)模型,设计了多模型自适应卡尔曼滤波器,并将多模型自适应卡尔曼滤波器应用于电动汽车电池荷电状态(state-of-charge,SOC)估计。由于老化电池是未知系统,利用传统的单一模型卡尔曼滤波器估计老化电池SOC时,因模型不准确而使估计误差增大。与单一模型滤波估计相比,多模型滤波估计融合了电池的各种老化信息,适合于未知系统的状态估计,从而提高了SOC的估计精度,并通过实验证明了上述结论的正确性。利用多模型自适应卡尔曼滤波器估计电池SOC,老化电池的模型与权值最大的单一模型较接近,根据单一模型权值可以近似估计出老化电池的健康状态(state of health,SOH),并通过电池容量测量,证明了SOH估计的正确性。

Based on the battery Thevenin model, the multiple-model adaptive kalman filter was proposed and applied to the electric vehicle battery state-of-charge （SOC） estimation. The ageing battery was an unknown system, when using the traditional single model kalman filter to estimate the SOC of ageing battery, the estimation error would be large induced by inaccuracy of the model. Compared with the single model filter estimation, the multiple-model filter was more suitable for an unknown system in the sense that it fused various ageing battery information, so this method could improve the estimation accuracy. The above conclusion is well proved by experiments. Using multiple-model adaptive kalman filter to estimate the battery SOC, the actual battery model is close to the model with the largest weight. In this regard, we can approximately estimate the battery state-of-health （SOH） according to the model Weight, which can be proved by the battery capacity measurement.