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  

SOC prediction of VRLA battery for HEV based on parameter-optimized least square support vector machines

作　　者：王琪[1,2] 孙玉坤[2,3] 黄永红[1,2] 

机构地区：[1]江苏大学电气信息工程学院,江苏镇江212013 [2]江苏大学机械工业设施农业测控技术与装备重点实验室,江苏镇江212013 [3]南京工程学院电力工程学院,江苏南京211167

出　　处：《中南大学学报（自然科学版）》2015年第1期113-119,共7页Journal of Central South University:Science and Technology

基　　金：国家自然科学基金资助项目(61074019;51377074);江苏高校优势学科建设工程资助项目(苏政办发[2011]6号);镇江市工业科技支撑项目(GY2010005);江苏省研究生创新计划项目(CXZZ13_0683)~~

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。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.

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 

分 类 号：TP912[自动化与计算机技术]