基于无迹粒子滤波的车载锂离子电池状态估计

State-of-Charge Estimation of Lithium-Ion Battery Using Unscented Particle Filter in Vehicle

传统的无迹卡尔曼滤波(UKF)和粒子滤波(PF)算法估计动力锂离子电池的荷电状态(SOC)时,常会出现电池模型参数不准确或粒子退化等问题导致估计精度差甚至系统发散等现象。为解决粒子匮乏和噪声干扰等问题,提出一种改进的估计算法——无迹粒子滤波算法(UPF)以实现SOC的精确估计。运用无迹卡尔曼算法为每个粒子计算均值和协方差,解决粒子滤波技术中粒子退化的问题。通过锂离子电池充放电实验,对等效模型进行辨识,最后在脉冲充放电和UDDS动态工况下对该算法进行测试验证。实验结果证明,基于二阶RC等效电路模型的UPF算法能显著提高SOC估计的实时性和精确性,其SOC估计精度在2%以内,收敛速度在250 s内。

The inaccurate battery models and particle degeneration problems often result in estimation errors or even divergence over time using the traditional unscented Kalman filter(UKF)and particle filter(PF)algorithms to estimate the state of charge(SOC)of power battery.In this study,an innovation method based on the unscented particle filter(UPF)is presented to suppress the particle degeneracy and noise interference.The unscented Kalman algorithm is used to calculate the mean and covariance for each particle and solve the problem of particle degeneration in particle filter technology.Through the lithium-ion battery charge-discharge test,the equivalent model is identified,and finally the algorithm is tested and verified under the pulse charge-discharge and UDDS dynamic conditions.The results show that the UPF method based on the two RC equivalent circuit model can improve the real-time performance and the precision of SOC estimation,and the estimation accuracy is less than 2%,the convergence rate is less than 250 s.