HEV电池SOC预测的留一交叉验证优化LS-SVM方法  

Battery SOC prediction for HEV based on LS-SVM within Leave-one-out cross-validation

作　　者：李可[1] 赵德安[1] 

机构地区：[1]江苏大学电气学院,江苏镇江212013

出　　处：《电源技术》2014年第11期2059-2062,共4页Chinese Journal of Power Sources

基　　金：江苏省研究生创新计划项目(CXLX13_0668)

摘　　要：针对混合动力汽车(HEV)电池剩余容量(SOC)判别问题,将最小二乘支持向量机方法应用于混合动力汽车电池荷电状态的预测。考虑到最小二乘支持向量机的参数选择会对预测结果产生较大的影响,提出了基于留一交叉验证优化最小二乘支持向量机的预测方法。将电池的工作电压、工作电流和表面温度参数用来预测蓄电池的荷电状态实时值,在欧洲城市行驶循环工况(EUDS)条件下进行实验验证,结果表明:所设计预测模型能够实时准确地预测出SOC值,有效性高。The least square support vector machines （LS-SVM） were proposed to predict the battery＇s state of charge （SOC） of the hybrid electric vehicles （HEV）. In consideration of the parameter selection of support vector machines exerts a major influence on SOC predict, a SOC prediction algorithm on the basis on Leave-one-out cross-validation （LOOCV） optimized least square support vector machines was presented. The working voltages, currents and surface temperature of battery were used to predict the real-time value of SOC. Experiment was verified under European Urban Driving Schedule （EUDS）. The results indicate that the prediction model possess higher predicted accuracy, achieve real-time and accurate SOC prediction.

关 键 词：SOC预测 留一交叉验证 最小二乘支持向量机 混合动力汽车 

分 类 号：TM912[电气工程—电力电子与电力传动]