基于支持向量机与改进高斯过程混合模型的车用电池容量预测方法*

doi:10.11985/2024.01.009

摘要/Abstract

摘要：

基于数据驱动的容量预测有助于锂电池健康管理以延长其使用寿命。然而，目前大多数相关方法基于实验室数据展开，无法反映实际复杂工况下车用电池老化特性。因此，本文利用电动汽车实车数据，设计了一种基于支持向量机与改进高斯过程的混合模型，实现了车用电池容量的精确预测。首先从汽车实时运行数据集中利用滑动窗口安时积分法提取其容量数据，设计了集合经验模态分解方法，将电池容量分为长期退化趋势和短期波动两部分，然后分别设计支持向量机与改进高斯过程对这两个分量进行建模，将结果融合得到最终的容量预测值。基于三辆实车数据集的试验结果表明，所提出的方法可以适用于实车数据的高精度容量预测。

关键词: 实车数据, 容量预测, 支持向量机, 改进高斯过程

Abstract:

The data-driven-based capacity prediction is essential for lithium-ion battery health management to extend its lifetime. However, most of the state-of-the-art methods are based on laboratory data analysis, which can not reflect the aging characteristics of the actual vehicle battery under complex conditions. Therefore, a hybrid model based on support vector machine and improved Gaussian process is designed to accurately predict the capacity of electric vehicle battery by using real vehicle data. Firstly, the capacity data is extracted from the real-time operation data set of vehicles by using the sliding window ampere-hour integration method, and the ensemble empirical mode decomposition method is designed to divide the battery capacity into two parts: long-term degradation trend and short-term fluctuation. Then, the support vector machine and the improved Gaussian process are designed respectively to model the two components, and the results are fused to obtain the final capacity prediction value. Extensive experiment results on three vehicles validate the effectiveness of the proposed hybrid prediction model, with higher accuracy demonstrated than the commonly used methods.

Key words: Real vehicle data, capacity prediction, support vector machine, improved Gaussian process

中图分类号:

TM912

李雨佳, 欧阳权, 刘灏仪, 祝铭烨, 王志胜. 基于支持向量机与改进高斯过程混合模型的车用电池容量预测方法^*[J]. 电气工程学报, 2024, 19(1): 87-96.

LI Yujia, OUYANG Quan, LIU Haoyi, ZHU Mingye, WANG Zhisheng. Vehicle Battery Capacity Prediction Based on Hybrid Model of Support Vector Machine and Improved Gaussian Process[J]. Journal of Electrical Engineering, 2024, 19(1): 87-96.

图/表 14

图1

图2

图3

图4

图5

图6

表1

图7

图8

图9

图10

表2

图11

表3

参考文献 20

[1]	赵振利. 电动汽车私桩共享充电定价模型及效益分配研究[D]. 北京: 华北电力大学, 2020.
	ZHAO Zhenli. Study on pricing model and benefit distribution of charging pile sharing for electric vehicles[D]. Beijing: North China Electric Power University, 2020.
[2]	KIM I L S. A technique for estimating the state of health of lithium batteries through a dual-sliding-mode observer[J]. IEEE Transactions on Power Electronics, 2009, 25(4):1013-1022. doi: 10.1109/TPEL.2009.2034966
[3]	耿星, 王友仁. 蓄电池SOH估算方法研究综述[J]. 机械制造与自动化, 2019, 48(1):210-212.
	GENG Xing, WANG Youren. Review of SOH estimation method for battery[J]. Mechanical Manufacturing and Automation, 2019, 48(1):210-212.
[4]	HASHEMI S R, MAHAJAN A M, FARHAD S. Online estimation of battery model parameters and state of health in electric and hybrid aircraft application[J]. Energy, 2021, 229(1):120699. doi: 10.1016/j.energy.2021.120699
[5]	ZHOU L, YANG Z, LI D, et al. State-of-health estimation for LiFePO₄ battery system on real-world electric vehicles considering aging stage[J]. IEEE Transactions on Transportation Electrification, 2021, 8(2):1724-1733. doi: 10.1109/TTE.2021.3129497
[6]	周頔, 宋显华, 卢文斌, 等. 基于日常片段充电数据的锂电池健康状态实时评估方法研究[J]. 中国电机工程学报, 2019, 39(1):107-113,327.
	ZHOU Di, SONG Xianhua, LU Wenbin, et al. Real-time SOH estimation algorithm for lithium-ion batteries based on daily segment charging data[J]. Proceedings of the CSEE, 2019, 39(1):107-113,327.
[7]	DRISCOLL L, De La TORRE S, GOMEZ-RUIZ J A. Feature-based lithium-ion battery state of health estimation with artificial neural networks[J]. Journal of Energy Storage, 2022,50:104584.
[8]	DUAN W, SONG S, XIAO F, et al. Battery SOH estimation and RUL prediction framework based on variable forgetting factor online sequential extreme learning machine and particle filter[J]. Journal of Energy Storage, 2023,65:107322.
[9]	张新锋, 姚蒙蒙, 王钟毅, 等. 基于ACO-BP神经网络的锂离子电池容量衰退预测[J]. 储能科学与技术, 2020, 9(1):138-144. doi: 10.19799/j.cnki.2095-4239.2019.0190
	ZHANG Xinfeng, YAO Mengmeng, WANG Zhongyi, et al. Lithium-ion battery capacity decline prediction based on ant colony optimization BP neural network algorithm[J]. Energy Storage Science and Technology, 2020, 9(1):138-144. doi: 10.19799/j.cnki.2095-4239.2019.0190
[10]	CHENG G, WANG X, HE Y. Remaining useful life and state of health prediction for lithium batteries based on empirical mode decomposition and a long and short memory neural network[J]. Energy, 2021,232:121022.
[11]	JIA J, LIANG J, SHI Y, et al. SOH and RUL prediction of lithium-ion batteries based on Gaussian process regression with indirect health indicators[J]. Energy, 2020, 13(2):375.
[12]	范智伟, 乔丹, 崔海港. 锂离子电池充放电倍率对容量衰减影响研究[J]. 电源技术, 2020, 44(3):325-329.
	FAN Zhiwei, QIAO Dan, CUI Haigang. Research on the effect of charge and discharge rates on capacity fading of lithium-ion batteries[J]. Journal of Power Sources, 2020, 44(3):325-329.
[13]	武佳卉, 邵振国, 杨少华, 等. 数据清洗在新能源功率预测中的研究综述和展望[J]. 电气技术, 2020, 21(11):1-6.
	WU Jiahui, SHAO Zhenguo, YANG Shaohua, et al. Review and prospect of data cleaning in renewable energy power prediction[J]. Electrical Engineering, 2020, 21(11):1-6.
[14]	胡杰, 何陈, 朱雪玲, 等. 基于实车数据的电动汽车电池剩余使用寿命预测[J]. 交通运输系统工程与信息, 2022, 22(1):292-300.
	HU Jie, HE Chen, ZHU Xueling, et al. Predicting remaining useful life of electric vehicle battery based on real vehicle data[J]. Journal of Transportation Systems Engineering and Information Technology, 2022, 22(1):292-300.
[15]	杨心月, 荆博, 梅志刚, 等. 风电机组功率异常数据剔除方法研究[J/OL]. 电测与仪表:1-9[2024-01-25]. http://kns.cnki.net/kcms/detail/23.1202.TH.20221025.1809.006.html.
	YANG Xinyue, JING Bo, MEI Zhigang, et al. Methods for elimination of abnormal power data of wind turbine[J/OL]. Electrical Measurement and Instrumentation:1-9[2024-01-25]. http://kns.cnki.net/kcms/detail/23.1202.TH.20221025.1809.006.html.
[16]	LI Q, LI R, JI K, et al. Kalman filter and its application[C]// 2015 8th International Conference on Intelligent Networks and Intelligent Systems (ICINIS),Tianjin,China,2015: 74-77.
[17]	WU Zhaohua, HUANG N E. Ensemble empirical mode decomposition:A noise-assisted data analysis method[J]. Advances in Adaptive Data Analysis, 2009, 1(1):1-41. doi: 10.1142/S1793536909000047
[18]	OTCHERE D A, GANAT T O A, GHOLAMI R, et al. Application of supervised machine learning paradigms in the prediction of petroleum reservoir properties:Comparative analysis of ANN and SVM models[J]. Journal of Petroleum Science and Engineering, 2021,200:108182.
[19]	LI L, WANG P, CHAO K-H, et al. Remaining useful life prediction for lithium-ion batteries based on Gaussian processes mixture[J]. PloS One, 2016, 11(9):e0163004. doi: 10.1371/journal.pone.0163004
[20]	ZHU M, OUYANG Q, WAN Y, et al. Remaining useful life prediction of lithium-ion batteries:A hybrid approach of grey-Markov chain model and improved Gaussian process[J]. IEEE Journal of Emerging and Selected Topics in Power Electronics, 2023, 11(1):143-153. doi: 10.1109/JESTPE.2021.3098378

数据集	模型	RMSE	MAE	MAPE(%)
实车1	混合模型	0.042 4	0.031 4	0.028 1
	SVM	0.147 8	0.116 0	0.104 4
	LSTM	0.421 3	0.398 5	0.357 8
	BP	0.076 7	0.066 4	0.064 5
实车2	混合模型	0.048 4	0.032 3	0.031 0
	SVM	0.100 9	0.083 9	0.081 0
	LSTM	0.198 9	0.161 0	0.155 5
	BP	0.064 4	0.051 9	0.034 6
实车3	混合模型	0.020 3	0.015 4	0.014 9
	SVM	0.075 7	0.056 9	0.055 2
	LSTM	0.019 8	0.016 7	0.016 2
	BP	0.098 5	0.072 0	0.057 3

预测步长	方法	实车1			实车2			实车3
预测步长	方法	RMSE	MAE	MAPE(%)	RMSE	MAE	MAPE(%)	RMSE	MAE	MAPE(%)
三步	融合方法	0.084 2	0.061 3	0.054 9	0.086 0	0.060 3	0.058 0	0.040 3	0.031 0	0.291 4
	SVM	0.169 8	0.133 9	0.120 5	0.124 9	0.100 1	0.096 7	0.103 7	0.076 7	0.335 7
	LSTM	0.390 7	0.313 7	0.280 0	0.337 5	0.288 7	0.278 6	0.071 1	0.059 4	0.057 6
	BP	0.128 5	0.113 7	0.119 1	0.164 2	0.136 3	0.105 3	0.130 5	0.104 0	0.096 2
六步	融合方法	0.148 5	0.094 6	0.084 7	0.181 4	0.117 2	0.115 2	0.076 2	0.055 6	0.053 9
	SVM	0.136 8	0.113 2	0.101 7	0.168 3	0.136 2	0.131 6	0.107 8	0.080 5	0.078 1
	LSTM	0.355 5	0.325 6	0.291 2	0.523 2	0.448 1	0.432 4	0.147 3	0.117 4	0.113 9
	BP	0.169 2	0.149 1	0.168 8	0.205 7	0.173 3	0.121 8	0.151 0	0.121 2	0.113 4

数据集	模型	RMSE	MAE	MAPE(%)
实车1	混合模型	0.042 6	0.032 3	0.028 8
	SVM	1.097 1	0.055 8	0.768 9
	LSTM	0.319 8	0.251 1	0.225 6
	BP	0.105 6	0.085 8	0.065 1
实车2	混合模型	0.050 5	0.033 0	0.085 8
	SVM	0.370 7	0.309 1	0.297 8
	LSTM	0.613 2	0.583 5	0.559 0
	BP	0.106 6	0.086 0	0.071 0
实车3	混合模型	0.018 9	0.014 3	0.013 9
	SVM	0.088 1	0.074 3	0.072 1
	LSTM	0.098 9	0.078 6	0.076 0
	BP	0.161 0	0.117 9	0.114 3