基于参数优化的最小二乘支持向量机HEV阀控铅酸蓄电池SOC预测

来源期刊:中南大学学报(自然科学版)2015年第1期

论文作者:王琪 孙玉坤 黄永红

文章页码:113 - 120

关键词:蓄电池;荷电状态;最小二乘支持向量机;参数优化;预测

Key words:battery; state of charge (SOC); least square support vector machines (LS-SVM); parameter-optimized; prediction

摘    要:针对电池容量预测问题,引入最小二乘支持向量机(LS-SVM)方法用于判断混合动力汽车(HEV)阀控铅酸蓄电池(VRLA)的荷电状态(SOC)。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,提出一种基于参数优化的最小二乘支持向量机预测方法。首先,在非线性回归预测模型的训练过程中,采用模拟退火算法来确定LS-SVM的初始值参数,从而更好地反映预测模型的复杂度,以此提高状态预测的精度。其次,由于预测模型在应对不良数据时可能出现误差增大的问题,分别采用贝叶斯证据框架(BEF)优化算法和留一交叉验证(LOOCV)优化算法来增强预测模型的抗差能力。研究结果表明:留一交叉验证优化算法具有较高的预测精度,实用性强,有效性高。

Abstract: Least square support vector machines (LS-SVM) method was used to predict valve regulated lead acid (VRLA) battery’s state of charge (SOC) for hybrid electric vehicles (HEV). Considering that parameter selection of support vector machines exerted a major influence on SOC predict, a SOC prediction algorithm was presented on the basis of parameter-optimized least square support vector machines. Firstly, in the training process of nonlinear regression predicted model, simulate anneal arithmetic was adopted to determine starting values of the LS-SVM algorithm for better reflecting the complexity of the predicted model and thus increased the predicted accuracy. Secondly, owing to that the predicted model maybe produces big errors, Bayesian evidence framework (BEF) and leave-one-out cross-validation (LOOCV) optimization methods were appended to enhance robustness of the predicted model by means of analyzing characteristic of kernel parameter. The results indicate that the leave-one-out cross-validation optimization method possessed higher predicted accuracy and effectiveness.

相关论文

  • 暂无!

相关知识点

  • 暂无!

有色金属在线官网  |   会议  |   在线投稿  |   购买纸书  |   科技图书馆

中南大学出版社 技术支持 版权声明   电话:0731-88830515 88830516   传真:0731-88710482   Email:administrator@cnnmol.com

互联网出版许可证:(署)网出证(京)字第342号   京ICP备17050991号-6      京公网安备11010802042557号