基于高斯混合回归的锂离子电池SOC估计

doi:10.19799/j.cnki.2095-4239.2019.0231

摘要/Abstract

摘要：

动力电池剩余电量的准确估算是电动汽车续航里程的重要依据和能量管理的基本前提，为降低电池系统因非线性，不平稳因素对荷电状态估计的不利影响。针对锂电池的状态数据采集存在非线性、不平稳以及外界干扰等问题，提出了基于高斯混合回归对荷电状态进行预测，以解决传统高斯过程模型含有异常数据和噪声的问题。利用K-means聚类算法与EM算法对高斯混合模型的超参数进行求解，然后采用高斯混合回归对输出的荷电状态进行预测。最后通过实验验证，并与高斯过程回归进行对比分析，验证了高斯混合回归算法在荷电状态估计过程中具有高精度和有效性。

关键词: 动力电池, 荷电状态, 高斯过程回归, 高斯混合回归

Abstract:

The state of charge (SOC) of a power battery must be accurately estimated as it determines the endurance mileage and is the basic premise for the energy management of electric vehicles. However, SOC estimation of battery systems is degraded by nonlinearity, instability, and other factors. Accordingly, the characteristic state data of a lithium battery contain nonlinearities, fluctuations, and external interference. This study proposes an SOC prediction method based on Gaussian mixture regression (GMR), which resolves the problems of abnormal values embedded in the state data and noise in the traditional Gaussian procession model (GPM). The hyper-parameters of the Gaussian mixture model are sequentially optimized by k-means clustering and an EM algorithm. The GMR predicts the SOC output. In an experimental validation and a comparative analysis of GMR and GPM, the GMR algorithm achieved superior prediction accuracy and effectiveness in SOC estimation.

Key words: power battery, state of charge, Gaussian process regression, Gaussian mixture regression

中图分类号:

TM 911

魏孟, 李嘉波, 叶敏, 高康平, 徐信芯. 基于高斯混合回归的锂离子电池SOC估计[J]. 储能科学与技术, 2020, 9(3): 958-963.

WEI Meng, LI Jiabo, YE Min, GAO Kangping, XU Xinxin. SOC estimation of Li-ion battery based on gaussian mixture regression[J]. Energy Storage Science and Technology, 2020, 9(3): 958-963.

图/表 10

表1

基于GMR的电池SOC估计"

算法：基于GMR的电池SOC估计
Step1.输入训练集数据
Step2.通过K-means进行聚类
Step3.建立电流，电压和SOC的联合密度GMM模型 $P (x) = ∑ k = 1 k α k N (μ k, Σ k)$
Step4.利用EM算法得到模型的最优参数 $Q = α 1, μ 1, Σ 1, α 2, μ 2, Σ 2, . . ., α n, μ n, Σ n$
Step5.输入测试数据，并计算相应高斯分量测试集的后验概率 $μ j x = μ j y + Σ j y x ? Σ j x x - 1 (x - μ j x)$ $Σ j x = Σ j y y - ?? Σ j y x ? Σ j x x - 1 Σ j x y$
Step6.根据GMR，输出预测结果 $μ x = ∑ j = 1 k α j x μ j x$ $ν x = ∑ j = 1 k α j x μ j x 2 + Σ j 2 - ?? ∑ j = 1 k α j x μ j x 2$

表1

图1

图2

图3

图4

图5

图6

图7

图8

表2

参考文献 18

1	CHENG C , RUI X , SHEN W . A lithium-ion battery-in-the-loop approach to test and validate multi-scale dual H infinity filters for state of charge and capacity estimation[J]. IEEE Transactions on Power Electronics, 2018(99): doi: 10.1109/TPEL.2017.2670081.
2	MENG J , LUO G , FEI G . Lithium polymer battery state-of-charge estimation based on adaptive unscented Kalman filter and support vector machine[J]. IEEE Transactions on Power Electronics, 2015, 31(3): 2226-2238.
3	ZHANG Y , RUI X , HE H , et al . Lithium-ion battery pack state of charge and state of energy estimation algorithms using a hardware-in-the-loop validation[J]. IEEE Transactions on Power Electronics, 2017, 32(6): 4421-4431.
4	SHENG H , XIAO J , WANG P . Lithium iron phosphate battery electric vehicle state-of-charge estimation based on evolutionary gaussian mixture regression[J]. IEEE Transactions on Industrial Electronics, 2016, 64(1): 544-551.
5	盛瀚民 .计及数据不可靠性的动力电池组SOC估计方法研究[D]. 成都: 西南交通大学, 2017.
	SHENG Hanmin . A study on power battery SOC estimation considering date uncertainty[D]. Chengdu: Southwest Jiatong University, 2017.
6	程晨, 顾伟, 褚建新, 等 . 基于Sage-Husa自适应滤波算法的锂电池荷电状态估计[J].南京工业大学学报(自然科学版), 2016, 38(3): 126-130.
	CHENG Chen , GU Wei , CHU Jianxin , et al . State of charge estimation of lithium batteries based on Sage-Husa adaptive filter algorithm[J]. Journal of Nanjing Tech University (Natural Science Edition), 2016, 38(3): 126-130.
7	田茂飞, 安治国, 陈星, 等 . 基于在线参数辨识和AEKF的锂电池SOC估计[J]. 储能科学与技术, 2019, 8(4): 745-750.
	TIAN Maofei , AN Zhiguo , CHEN Xing , et al . SOC estimation of lithium battery based online parameter identification and AEKF[J]. Energy Storage Science and Technology, 2019, 8(4): 745-750.
8	LI B , BEI S . Estimation algorithm research for lithium battery SOC in electric vehicles based on adaptive unscented Kalman filter[J]. Neural Computing and Applications, 2018： doi: 10.1007/s00521-018-3901-7.
9	YUE L , CHATTOPADHYAY P , XIONG S , et al . Dynamic data-driven and model-based recursive analysis for estimation of battery state-of-charge[J]. Applied Energy, 2016, 184: 266-275.
10	苏振浩, 李晓杰, 秦晋, 等 . 基于BP人工神经网络模型的动力电池SOC估计方法[J]. 储能科学与技术, 2019, 8(5): 868-873.
	SU Zhenhao , LI Xiaojie , QIN Jin , et al . SOC estimation method of power battery based on BP artificial neural network[J]. Energy Storage Science and Technology, 2019, 8(5): 868-873.
11	戴庚, 耿诗尧 .基于最小二乘支持向量机的SOC估计方法[J]. 信息与电脑(理论版), 2018, 39(2): 36-37.
	DAI Geng , GENG Shiyao . SOC estimation based on least squares support vector machines[J].China Computer & Communication, 2018, 39(2): 36-37.
12	MEI C , SU Y , LIU G , et al . Dynamic soft sensor development based on Gaussian mixture regression for fermentation processes[J]. Chinese Journal of Chemical Engineering, 2016, 25(1): 116-122.
13	SAHINOGLU G O , PAJOVIC M , SAHINOGLU Z , et al . Battery state of charge estimation based on regular/recurrent gaussian process regression[J]. IEEE Transactions on Industrial Electronics, 2017(99): doi: 10.1109/TIE.2017.2764869.
14	LIU D , PANG J , ZHOU J , et al . Prognostics for state of health estimation of lithium-ion batteries based on combination Gaussian process functional regression[J]. Microelectronics Reliability, 2013, 53(6): 832-839.
15	何志昆, 刘光斌, 赵曦晶, 等 . 高斯过程回归方法综述[J]. 控制与决策, 2013(8): 1121-1129.
	HE Zhikun , LI Guangbin , ZHAO Xijing , et al . Overview of Gaussian process regression [J]. Control and Decision, 2013(8): 1121-1129.
16	李嘉波, 魏孟, 叶敏, 等 . 基于高斯过程回归的锂离子电池SOC估计[J]. 储能科学与技术, 2019， 9(1): 131-137.
	LI Jiabo , WEI Meng , YE Min , et al . SOC estimation of lithium-ion batteries based on Gauss process regression[J]. Energy Storage Science and Technology, 2019， 9(1): 131-137.
17	周亚同, 赵翔宇, 何峰, 等 . 基于高斯过程混合模型的大气温湿度预测[J]. 农业工程学报, 2018， 34(5): 219-226.
	ZHOU Yatong , ZHAO Xiangyu , HE Feng , et al . Atmospheric temperature and humidity prediction of Gaussian process mixed model[J]. Transactions of the Chinese Society of Agricultural Engineering, 2018， 34(5): 219-226.
18	杨飞, 乔铁柱, 庞宇松, 等 . 基于改进的高斯混合回归的球磨机料位软测量[J]. 现代电子技术, 2018， 41(5): 153-158.
	YANG Fei , QIAO Tiezhu , PANG Yusong , et al . Soft measurement for ball mill fill level based on improved Gaussian mixture regression[J]. Modern Electronics Technique, 2018， 41(5): 153-158.

方法	RMSE/%	MAPE/%
GPR（UDDS）	3.34	3.79
GMR（UDDS）	2.37	2.23
GPR（DST）	0.91	2.24
GMR（DST）	0.64	2.22

[1]	吴田, 林闽城, 海浩, 孙海渔, 温兆银, 马福元. 面向一次调频的镍氢电池系统开发[J]. 储能科学与技术, 2022, 11(7): 2213-2221.
[2]	常泽宇, 张之琦, 张晓东, 李丽, 郁亚娟. 基于数据驱动的动力电池健康状态评估平台[J]. 储能科学与技术, 2022, 11(6): 1847-1853.
[3]	周伟, 符冬菊, 刘伟峰, 陈建军, 胡照, 曾燮榕. 废旧磷酸铁锂动力电池回收利用研究进展[J]. 储能科学与技术, 2022, 11(6): 1854-1864.
[4]	田峰, 程志江, 杨涵棣, 杨天翔. 模块化多电平复合变换器电池储能系统容错控制策略[J]. 储能科学与技术, 2022, 11(5): 1583-1591.
[5]	王苏杭, 李建林, 李雅欣, 熊俊杰, 曾伟. 锂离子电池系统低温充电策略[J]. 储能科学与技术, 2022, 11(5): 1537-1542.
[6]	张孝远, 张金浩, 蒋亚俊. 基于改进TCN模型的动力电池健康状态评估[J]. 储能科学与技术, 2022, 11(5): 1617-1626.
[7]	王芳, 王峥, 林春景, 张国振, 张贵平, 马天翼, 刘磊, 刘仕强. 新能源汽车动力电池安全失效潜在原因分析[J]. 储能科学与技术, 2022, 11(5): 1411-1418.
[8]	刘春辉, 任宏斌. 基于SOC的动力电池组主动均衡研究[J]. 储能科学与技术, 2022, 11(2): 667-672.
[9]	黄鹏超, 鄂加强. 基于双自适应卡尔曼滤波的锂电池状态估算[J]. 储能科学与技术, 2022, 11(2): 660-666.
[10]	王帅, 马鸿雁, 窦嘉铭, 张英达, 李晟延, 胡璐锦. 基于UGOA-BP的锂电池SOC估算[J]. 储能科学与技术, 2022, 11(1): 258-264.
[11]	高晓芝, 王磊, 田晋, 刘佳璐, 刘庆华. 基于参数优化变分模态分解的混合储能功率分配策略[J]. 储能科学与技术, 2022, 11(1): 147-155.
[12]	孙爱芬, 赤娜. 基于改进的高斯过程回归的SOC估计算法[J]. 储能科学与技术, 2022, 11(1): 253-257.
[13]	马天翼, 王伟哲, 林春景, 白广利, 韦振, 刘仕强. 磷酸铁锂电池多阶段利用可行性[J]. 储能科学与技术, 2022, 11(1): 119-126.
[14]	胡建, 林春景, 郝维健, 郑天雷. 动力电池标准体系建设现状及建议[J]. 储能科学与技术, 2022, 11(1): 313-320.
[15]	曹新宇, 彭飞, 李立伟, 尹建光. 基于IBAS-NARX神经网络的锂电池荷电状态估计[J]. 储能科学与技术, 2021, 10(6): 2342-2351.