【摘要】 利用扩展卡尔曼滤波算法对磷酸铁锂电池进行SOC估算时,系统噪声和观测噪声的噪声协方差矩阵多为随机给出,无法对噪声问题进行针对性的优化。基于上述问题,提出了一种基于IPSO-EKF的融合算法,在动态工况下优化噪声协方差矩阵,提高SOC估算精度。试验和仿真结果表明:相对于EKF算法,所提出的IPSO-EKF算法在准确性和适应范围上有更好的表现;收敛速度较快,在5次左右的迭代过程中迅速收敛到全局最优位置,并且在随后的迭代过程中,最佳适应度值趋向于稳定;通过RMSE和MAPE值评价算法的可靠性,在DST、UDDS及NEDC工况下,RMSE值分别为0.224 4、0.198 0和0.368 4,MAPE值分别为0.605 0、0.668 0和0.706 7。此外,还提供了通过噪声寻优提高SOC估算精度的思路。更多还原

【Abstract】 When the extended Kalman filter algorithm is used to estimate the SOC of lithium iron phosphate battery, there is a serious problem, that is, the noise covariance matrix of system noise and observation noise is mostly given randomly, so it is impossible to optimize the noise problem. To address this problem, this paper proposes a fusion algorithm combining EKF(Extended Kalman Filter) and IPSO(Improved Particle Swarm Optimization), namely IPSO-EKF, to optimize noise covariance under variant working conditions to improve the accuracy of SOC estimation. The results of experiment and simulation indicate that the proposed method has an edge on SOC estimation in accuracy and adaptability. Besides, the convergence speed is fast, which converges to the global optimal position in about 5 iterations and fluctuates to be stable in the subsequent iterations. Furthermore, the reliability of the IPSO-EKF algorithm is evaluated through RMSE and MAPE, and the values of former are 0.224 4, 0.198 0 and 0.368 4, as well as the values of latter are 0.605 0, 0.668 0 and 0.706 7 under DST, UDDS and NEDC profiles. In addition, an idea of improving SOC estimation accuracy through noise optimization is also provided.更多还原