劉榮 童亮 許永紅
摘 要: 針對(duì)混合動(dòng)力汽車(chē)在復(fù)雜工況下動(dòng)力電池溫度測(cè)量可靠性下降的問(wèn)題,提出基于pso_FSVM的車(chē)用動(dòng)力電池溫度預(yù)測(cè)模型,該研究分別采集車(chē)輛key_on和key_off兩種狀態(tài)下的動(dòng)力電池溫度數(shù)據(jù),采用粒子群優(yōu)化的快速支持向量機(jī)算法,構(gòu)建了穩(wěn)定的動(dòng)力電池溫度預(yù)測(cè)模型。實(shí)驗(yàn)結(jié)果表明,在車(chē)輛key_on和key_off兩種狀態(tài)下,數(shù)據(jù)集的預(yù)測(cè)數(shù)據(jù)與實(shí)際測(cè)量數(shù)據(jù)的相關(guān)系數(shù)分別達(dá)到0.810 2和0.797 3,溫度預(yù)測(cè)誤差小于2 ℃,pso_FSVM模型提高了動(dòng)力電池溫度預(yù)測(cè)的精度和可靠性。
關(guān)鍵詞: 混合動(dòng)力汽車(chē); 動(dòng)力電池溫度; 粒子群; 快速支持向量機(jī); 預(yù)測(cè)模型; 熱動(dòng)力學(xué)模型
中圖分類(lèi)號(hào): TN245?34; TP336 文獻(xiàn)標(biāo)識(shí)碼: A 文章編號(hào): 1004?373X(2018)12?0024?04
Abstract: In allusion to the problem of the decline of temperature measurement reliability for power battery of the hybrid electric vehicle in complicated working conditions, the temperature data of power battery at two vehicle states of Key_on and Key_off is collected respectively. A stable power battery temperature prediction model is constructed by using the particle swarm optimization based fast support vector machine algorithm. The experimental results show that the correlation coefficient between the prediction data and actual measurement data of data sets reaches 0.810 2 and 0.797 3 respectively at the two vehicle states of Key_on and Key_off, and the temperature prediction error is less than 2 ℃, which indicates that the pso_FSVM model can improve the prediction accuracy and reliability of power battery temperature.
Keywords: pso_FSVM; hybrid electric vehicle; power battery temperature; particle swarm optimization; fast support vector machine; prediction model; thermodynamics model
0 引 言
動(dòng)力電池是混合動(dòng)力汽車(chē)驅(qū)動(dòng)系統(tǒng)中必不可少的能源裝置,確保動(dòng)力電池的性能和使用壽命尤為重要。在復(fù)雜的工況下,電池溫度作為影響電池性能和使用壽命的關(guān)鍵因素之一,必須準(zhǔn)確地預(yù)測(cè)電池溫度以確保電池系統(tǒng)工作在最好狀態(tài)下。因此,構(gòu)建準(zhǔn)確可靠的電池溫度預(yù)測(cè)模型,實(shí)時(shí)在線地監(jiān)測(cè)電池溫度,為動(dòng)力電池及其管理系統(tǒng)保駕護(hù)航,對(duì)混合動(dòng)力汽車(chē)安全性能和動(dòng)力電池?zé)峁芾硐到y(tǒng)的研究設(shè)計(jì)同樣具有重大意義。近年來(lái)對(duì)電池性能的大量研究取得了重要成果,其中Hu借助Matlab/Simulink基于神經(jīng)網(wǎng)絡(luò)對(duì)PEMFC的搭建高溫穩(wěn)態(tài)性能和動(dòng)態(tài)性能的預(yù)測(cè)模型,預(yù)測(cè)精度較高[1]。Sun提出基于熱平衡方程的電池溫度在線方法[2]。Hong提出了一種基于電阻層析成像的動(dòng)力電池內(nèi)部溫度監(jiān)測(cè)新方法[3]。但是這些研究方法建立的模型過(guò)于復(fù)雜,也沒(méi)有建立在動(dòng)力電池處于復(fù)雜的工作工況下。本文在電池?zé)釀?dòng)力學(xué)模型理論基礎(chǔ)上,結(jié)合在車(chē)輛復(fù)雜工況下采集的測(cè)試數(shù)據(jù)集,提出一種基于pso_FSVM模型的電池溫度預(yù)測(cè)模型,pso_FSVM模型能夠準(zhǔn)確可靠地預(yù)測(cè)動(dòng)力電池溫度,為往后更為復(fù)雜的工況和環(huán)境(極寒、極熱)下預(yù)測(cè)電池溫度提供可行性。
1 電池?zé)釀?dòng)力學(xué)模型
電池表面溫度和內(nèi)部溫度是表征電池性能的兩個(gè)重要參數(shù),對(duì)于廣泛應(yīng)用于混合動(dòng)力汽車(chē)的動(dòng)力電池而言,內(nèi)部溫度是影響電池性能的決定性因素。在研究過(guò)程中,美國(guó)伯克利大學(xué)的D.Bernardi用可逆熵變產(chǎn)生的熱量描述化學(xué)反應(yīng)熱,提出了Bernardi 生熱率計(jì)算模型[4];而劉偉在研究中建立一種基于遞歸最小二乘法估算的電池?zé)釀?dòng)力學(xué)模型[5]。在車(chē)輛key_on和key_off狀態(tài)下,依據(jù)物理傳熱定律可得動(dòng)力電池?zé)釀?dòng)力學(xué)模型,如下:
4 結(jié) 論
1) 依據(jù)電池?zé)釀?dòng)力學(xué)數(shù)學(xué)模型,本文提出的粒子群優(yōu)化算法的快速支持向量機(jī)模型在對(duì)動(dòng)力電池溫度預(yù)測(cè)中運(yùn)算速度快,相對(duì)誤差極小,為在更為復(fù)雜的工況下提供研究?jī)r(jià)值,并且可以運(yùn)用于混合動(dòng)力汽車(chē)/純電動(dòng)汽車(chē)電池?zé)崃抗芾硐到y(tǒng)對(duì)熱量的最優(yōu)控制。
2) 在車(chē)輛全工況下,采集原始數(shù)據(jù)過(guò)程中應(yīng)考慮電池均衡帶來(lái)的影響,對(duì)大樣本數(shù)據(jù)的綜合處理以及粒子位置和速度的更新策略進(jìn)行更為合理的規(guī)劃,提高在大樣本情況下粒子群算法的尋優(yōu)能力。
注:本文通訊作者為童亮。
參考文獻(xiàn)
[1] 胡經(jīng)緯.質(zhì)子交換膜燃料電池的電化學(xué)和數(shù)值模擬研究[D].大連:中國(guó)科學(xué)院,2006.
HU Jingwei. Electrochemical and mathematical model studies on PEMFC [D]. Dalian: Chinese Academy of Sciences, 2006.
[2] 孫金磊,朱春波,李磊,等.電動(dòng)汽車(chē)動(dòng)力電池溫度在線估計(jì)方法[J].電工技術(shù)學(xué)報(bào),2017,32(7):197?203.
SUN Jinlei, ZHU Chunbo, LI Lei, et al. Online temperature estimation method for electric vehicle power battery [J]. Transactions of China Electrotechnical Society, 2017, 32(7): 197?203.
[3] 洪曉斌,李年智,尹文偉,等.基于電阻層析成像的汽車(chē)動(dòng)力電池內(nèi)部溫度監(jiān)測(cè)[J].光學(xué)精密工程,2014,22(1):193?203.
HONG Xiaobin, LI Nianzhi, YIN Wenwei, et al. Monitoring of internal temperature of vehicle power battery based on electrical resistance tomography [J]. Optics and precision engineering, 2014, 22(1): 193?203.
[4] BERNARDI D, PAWLIKOWSKI E, NEWMAN J. A general energy balance for battery systems [J]. Journal of Electrochemical Society, 1985, 132(1): 5?12.
[5] LIU Wei. Introduction to hybrid vehicle system modeling and control [M]. Hoboken: Wiley, 2013.
[6] KENNEDY J, EBERHART R. Particle swarm optimization [C]// Proceedings of IEEE International Conference on Neural Networks. Piscataway: IEEE, 1995: 1942?1948.
[7] 王建國(guó),張文興.支持向量機(jī)建模及其智能優(yōu)化[M].北京:清華大學(xué)出版社,2015:134?135.
WANG Jianguo, ZHANG Wenxing. Support vector machine modeling and intelligent optimization [M]. Beijing: Tsinghua University Press, 2015: 134?135.
[8] 彭宇,彭喜元,劉兆慶.微粒群算法參數(shù)效能的統(tǒng)計(jì)分析[J].電子學(xué)報(bào),2004,32(2):209?213.
PENG Yu, PENG Xiyuan, LIU Zhaoqing. Statistic analysis on parameter efficiency of particle swarm optimization [J]. Acta electronica sinica, 2004, 32(2): 209?213.
[9] 張文興,叢寬,王建國(guó),等.一種新的快速支持向量回歸算法[J].微計(jì)算機(jī)信息,2010,26(33):208?209.
ZHANG Wenxing, CONG Kuan, WANG Jianguo, et al. A new algorithm of fast support vector regression [J]. Microcomputer information, 2010, 26(33): 208?209.
[10] HUANG Mingzhi, HAN Wei, WAN Jinquan, et a1. Multi?objective optimization for design and operation of anaerobic digestion using GA?ANN and NSGA?II [J]. Journal of chemical technology and biotechnology, 2016, 91(1): 226?233.