一种新型锂电池充电剩余时间预测方法
|
摘要
提出了一种基于信息粒化(IG)的支持向量回归(SVR)方法来预测锂电池充电剩余时间。首先,通过模糊信息粒化窗口提取代表性数据,同时形成概率性预测的置信区间上下限,并重新组合特征向量建立训练样本。然后,运用样本对训练支持向量回归模型,在参数优选方面采用网格划分的交叉验证方式。最后,通过3个不同的支持向量回归模型得到充电剩余时间的置信区间。以美国国家航空航天阿姆斯研究中心公开的电池数据为实例,通过与三段式模型方法进行对比,结果表明该模型在精度、通用性方面表现更好。
A method for predicting the remaining charging time of a lithium-ion was proposed by using thecombination of information granulation(IG) and support vector regression(SVR). First, the windows of fuzzyinformation granulation were utilized to extract representative data. At the same time, the boundary of the confidenceinterval was obtained and the feature vectors were re-established to form training samples. Then the samples wereutilized to train the model of support vector regression. In order to get optimal parameters, an approach namedcross-validation based on mesh was used. Finally, the confidence interval of remaining charging time could be easilyobtained by three different models of the support vector regression. Based on the public battery data sets provided byNationl Aeronautics and Space Administration Ames Research Center, comparative experiments were conductedbetween the SVR with IG and three-stage model. The results demonstrate that the SVR with IG performs better interms of the accuracy and versatility.
A method for predicting the remaining charging time of a lithium-ion was proposed by using thecombination of information granulation(IG) and support vector regression(SVR). First, the windows of fuzzyinformation granulation were utilized to extract representative data. At the same time, the boundary of the confidenceinterval was obtained and the feature vectors were re-established to form training samples. Then the samples wereutilized to train the model of support vector regression. In order to get optimal parameters, an approach namedcross-validation based on mesh was used. Finally, the confidence interval of remaining charging time could be easilyobtained by three different models of the support vector regression. Based on the public battery data sets provided byNationl Aeronautics and Space Administration Ames Research Center, comparative experiments were conductedbetween the SVR with IG and three-stage model. The results demonstrate that the SVR with IG performs better interms of the accuracy and versatility.
引文
[1] GOEBEL K, SAHA B, SAXENA A, et al. Prognostics in batteryhealth management[J]. IEEE Instrumentation&Measurement Maga-zine, 2008, 11(4):33-40.
[2] LIU D, PANG J, ZHOU J, et al. Prognostics for state of health esti-mation of lithium-ion batteries based on combination Gaussian pro-cess functional regression[J]. Microelectronics Reliability, 2013, 53(6):832-839.
[3] CHAOUI H, GOLBON N, HMOUZ I, et al. Lyapunov-based adap-tive state of charge and state of health estimation for lithium-ion bat-teries[J]. IEEE Transactions on Industrial Electronics, 2015, 62(3):1610-1618.
[4]娄宇成,解大,房新雨,等.基于多元参数融合曲面的电池充电特性分析[J].太阳能学报,2015,36(4):823-828.
[5]周祖德,徐超.全CMOS三段式锂电池充电器设计[J].武汉理工大学学报,2006,28(4):115-117,121.
[6]赵弈昕,翁志祁.锂电池充电剩余时间的三段式模型[J].华岗工程学报,2010,26:73-82.
[7]韩涛,徐鸣,卢云飞.充电剩余时间估计方法、装置及移动设备:中国,CN102945200A[P].2013-02-27.
[8] ANSARI H R, GHOLAMI A.An improved support vector regressionmodel for estimation of saturation pressure of crude oils[J]. FluidPhase Equilibria, 2015, 402(3):124-132.
[9]陈雄姿,于劲松,唐荻音,等.基于贝叶斯LS-SVR的锂电池剩余寿命概率性预测[J].航空学报,2013,34(9):2219-2229.
[10] PEDRYCZ W, SUCCI G, SILLITTI A, et al. Data description:ageneral framework of information granules[J]. Knowledge-BasedSystems, 2015, 80:98-108.
[11] MENCAR C, FANELLI A M. Interpretability constraints for fuzzyinformation granulation[J]. Information Sciences, 2008, 178(24):4585-4618.
[2] LIU D, PANG J, ZHOU J, et al. Prognostics for state of health esti-mation of lithium-ion batteries based on combination Gaussian pro-cess functional regression[J]. Microelectronics Reliability, 2013, 53(6):832-839.
[3] CHAOUI H, GOLBON N, HMOUZ I, et al. Lyapunov-based adap-tive state of charge and state of health estimation for lithium-ion bat-teries[J]. IEEE Transactions on Industrial Electronics, 2015, 62(3):1610-1618.
[4]娄宇成,解大,房新雨,等.基于多元参数融合曲面的电池充电特性分析[J].太阳能学报,2015,36(4):823-828.
[5]周祖德,徐超.全CMOS三段式锂电池充电器设计[J].武汉理工大学学报,2006,28(4):115-117,121.
[6]赵弈昕,翁志祁.锂电池充电剩余时间的三段式模型[J].华岗工程学报,2010,26:73-82.
[7]韩涛,徐鸣,卢云飞.充电剩余时间估计方法、装置及移动设备:中国,CN102945200A[P].2013-02-27.
[8] ANSARI H R, GHOLAMI A.An improved support vector regressionmodel for estimation of saturation pressure of crude oils[J]. FluidPhase Equilibria, 2015, 402(3):124-132.
[9]陈雄姿,于劲松,唐荻音,等.基于贝叶斯LS-SVR的锂电池剩余寿命概率性预测[J].航空学报,2013,34(9):2219-2229.
[10] PEDRYCZ W, SUCCI G, SILLITTI A, et al. Data description:ageneral framework of information granules[J]. Knowledge-BasedSystems, 2015, 80:98-108.
[11] MENCAR C, FANELLI A M. Interpretability constraints for fuzzyinformation granulation[J]. Information Sciences, 2008, 178(24):4585-4618.