点击次数：

所属单位：能源与动力学院

发表刊物：Qiche Gongcheng

摘要：Accurate estimation of battery SOC is not only a prerequisite for control and optimization of electric vehicle, but also the basis for the reasonable implementation of battery management. In this paper, based on the first-order RC equivalent circuit model, a SOC estimation method for lithium power battery is proposed by combining BP neural network with extended Kalman filtering (EKF), in which recursive least-squares with forgetting factor and BP-EKF algorithm are adopted to conduct a co-estimation of SOC and other model parameters. Corresponding filter output parameters are used to train BP neural network off-line, and succed trained BP neural network are used to compensate the estimation error in EKF algorithm. The results of simulation and dynamic state test show that compared with EKF algorithm, the SOC estimation method proposed has good performances in divergence suppression and robustness, and can effectively enhance the estimation accuracy of battery SOC. © 2017, Society of Automotive Engineers of China. All right reserved.

ISSN号：1000-680X

是否译文：否

发表时间：2017-06-25

合写作者：孔祥创,王春燕

通讯作者：孔祥创,赵万忠