基于EKF-Markov方法的动力电池SOC预测  

Prediction of power battery state of charge based on GM-Markov

作　　者：潘盛辉[1] 胡三丽 郭毅锋[1] 韩峻峰[2] 

机构地区：[1]广西科技大学广西汽车零部件与整车技术重点实验室,广西柳州545006 [2]钦州学院,广西钦州535000

出　　处：《电源技术》2016年第5期990-993,共4页Chinese Journal of Power Sources

基　　金：国家自然科学基金项目(51407038);广西自然科学基金项目(2013GXNSFBA019241);广西汽车零部件与整车技术重点实验室建设项目课题(14-A-02-03);广西汽车零部件与整车技术重点实验室开放课题(2013KFMS02)

摘　　要：针对工况环境下动力电池SOC的变化具有非线性的特点,对未来SOC状态进行精确预测。首先采用EKF预测模型对动力电池SOC值进行预测,根据预测结果划分SOC状态区间,进一步得到SOC值的Markov状态转移矩阵,然后将EKF模型与Markov状态转移矩阵相结合对SOC进行预测。设计了UDDS工况下的实验验证方案来获取动力电池SOC数据样本,对比分析表明,EKF-Markov方法能够有效地削弱EKF方法所产生的预测误差累积效应,平均预测误差相较EKF降低了83.3%,可对动力电池SOC做出更精确的预测。Aiming at the non-linear character of the change of SOC of power battery which operated at working conditions. In order to make reasonably accurate prediction of SOC at some indeterminate point in the future, firstly,through the application of EKF model for predicting the SOC of power battery, basised on the results of prediction,dividing the state interval of SOC, state transfer matrix of Markov of SOC can be got futher, then combing with EKF model and state transfer matrix of Markov for predicting the SOC. The data sample of the SOC of power battery was presented by designing the experimental validation scheme in the working condition of UDDS. A comparative analysis revealed that the predictive error of the approach of EKF-Markov is reduced 83.3% than EKF,which can make relatively accurate prediction for the SOC of power battery, which can make more exact prediction of SOC.

关 键 词：动力电池 SOC EKF-Markov 预测 

分 类 号：TM912[电气工程—电力电子与电力传动]