采用改进最大相关熵自适应迭代容积卡尔曼滤波算法的锂离子电池荷电状态估计

Estimation of Lithium-Ion Battery State of Charge Using an Innovation Maximum Correlation-Entropy Criterion Adaptive Iterative Cubature Kalman Filter Algorithm

针对非高斯噪声干扰下传统滤波算法在估计锂离子电池荷电状态(SOC)时存在不稳定以及精度低的问题,提出一种改进的最大相关熵自适应迭代容积卡尔曼滤波(IMCC-AICKF)算法,用于估计锂离子电池荷电状态。所提算法将加权最小二乘方法与最大相关熵准则(MCC)相结合,定义了一种新的代价权函数作为优化准则,通过优化噪声最小协方差矩阵来减小滤波误差,保证长时间滤波的收敛性和稳定性;再与自适应迭代容积卡尔曼滤波(AICKF)算法相结合,对过程噪声协方差和测量噪声协方差进行更新来提高估计的准确性和鲁棒性。基于两种电池数据,在非高斯噪声干扰下,运用所提算法对电池SOC进行估计,仿真结果表明:与容积卡尔曼滤波(CKF)算法和最大相关熵容积卡尔曼滤波(IMCC-CKF)算法相比,IMCC-AICKF算法对荷电状态估计的最大绝对误差、平均绝对误差和均方根误差都是最小的,且平均绝对误差和均方根误差均小于1%;在给定初始值错误的情况下,IMCC-AICKF算法可以准确收敛到真实值,具有较好的鲁棒性。所提算法在非高斯噪声下能实现更准确的估计,是一种估计精度高且鲁棒性好的SOC估计方法。

In response to the issues of instability and low accuracy in estimating the state of charge(SOC)of lithium-ion batteries under non-Gaussian noise interference traditional filtering algorithms,an innovation maximum correlation-entropy criterion adaptive iterated cubature Kalman filtering algorithm(IMCC-AICKF)is proposed for SOC estimation of lithium-ion batteries.The proposed algorithm combines the weighted least squares method with the maximum correlation-entropy criterion(MCC)to define a new cost-weight function as the optimization criterion.This approach aids in reducing filtering errors by optimizing the minimum noise covariance matrix to reduce filtering errors and stability of long-term filtering.Subsequently,by integrating with the adaptive iterative covariance Kalman filter(AICKF),the process noise covariances and measurement noise covariances are updated to enhance estimation accuracy and robustness.Based on two sets of battery data and under non-Gaussian noise interference,the proposed algorithm is applied to estimate the SOC of the batteries.The simulation results demonstrate that compared to cubature Kalman filtering(CKF)and innovation maximum correlation-entropy criterion cubature Kalman filtering(IMCC-CKF),the IMCC-AICKF algorithm yields the smallest maximum absolute error(MaxAE),mean absolute error(MAE),and root mean square error(RMSE)in SOC estimation,with both MAE and RMSE below 1%.Additionally,even with initial value errors,IMCC-AICKF can accurately converge to the true values,demonstrating good robustness.The proposed algorithm achieves more accurate estimation under non-Gaussian noise,providing a high-precision and robust method for SOC estimation.