基于奇异值分解无迹卡尔曼滤波的锂电池荷电状态估计

锂电池的荷电状态(state of charge, SOC)是电池管理系统(battery management system, BMS)对锂电池进行管理的重要指标。针对传统SOC估计方法存在的精度低、计算复杂和鲁棒性差等问题，本文提出了一种基于奇异值分解无迹卡尔曼滤波(singular value decomposition unscented Kalman filter, SVD-UKF)的SOC估计方法。该方法利用无迹变换(unscented transformation，UT)提高了计算精度的同时降低了计算量，并且克服了UKF在状态协方差矩阵P非半正定时会出现滤波发散的缺点，提高了算法的鲁棒性。实验结果表明，该算法能够快速收敛于真值，并且将估算误差降低至1%。

Lithium Battery SOC Estimation Based on Singular Value Decomposition Unscented Kalman filter

The state of charge (SOC) of lithium batteries was an important indicator for the management of lithium batteries by the battery management system (BMS). Aiming at the problems of low precision, computational complexity and poor robustness of traditional SOC estimation methods, a SOC estimation method based on singular value decomposition unscented Kalman filter (SVD-UKF) was proposed. Unscented transformation (UT) was used by this method to improve the calculation accuracy and reduce the calculation amount. The shortcomings of the filter divergence of the UKF when the state covariance matrix P is not half-positive was eliminated. Also the robustness of the algorithm was improved. Experimental results show that the algorithm can quickly converge to the true value and reduce the estimation error to 1%.