基于交互式多模型无迹卡尔曼滤波的锂电池荷电状态估计

为了提高传统卡尔曼滤波法估计锂电池荷电状态(SOC)的精度,在锂电池二阶RC等效电路模型基础上,根据隐马尔科夫模型(HMM)理论并采用遗传算法优化构造出了不同参数状态的电池模型。结合交互式多模型(IMM)算法与无迹卡尔曼滤波(UKF)算法进行SOC估计,提出了一种基于HMM的IMM-UKF算法估计锂电池SOC的方法。锂电池在线SOC估计实验表明,该方法比较其他估计方法有着更高的估计精度,平均绝对误差仅为1%。

Estimation of State-of-Charge for Lithium-ion Battery by Interactive Multiple Model

In order to improve the accuracy of traditional Kalman filter method in estimating the state of charge (SOC) of lithium-ion battery, based on the second-order RC equivalent circuit model of lithium-ion battery, a battery model with different parameter states were constructed according to Hidden Markov Model (HMM) theory and optimized by genetic algorithm. Combining the Interactive Multiple Model (IMM) algorithm with unscented Kalman filter (UKF) algorithm for SOC estimation, an estimation method of SOC for lithium-ion battery based on IMM-UKF algorithm was proposed. The on-line SOC estimation experiment of lithium batteries shows that this method has higher estimation accuracy than other estimation methods, and the average absolute error is only 1%.