Please wait a minute...
J4  2012, Vol. 46 Issue (4): 705-711    DOI: 10.3785/j.issn.1008-973X.2012.04.019
Robust recursive kernel learning modeling method with
application to blast furnace
YU Hai-qing1, LIU Yi2, CHEN Kun1, JI Jun1, LI Ping1
1. State Key Laboratory of Industrial Control Technology, Zhejiang University, Hangzhou 310027, China; 2. Institute of
Process Equipment and Control Engineering, Zhejiang University of Technology, Hangzhou 310032, China
Download:   PDF(0KB) HTML
Export: BibTeX | EndNote (RIS)      


A robust recursive least squares support vector regression (R-RLSSVR) soft-sensing modeling method was proposed in order to overcome the embarrassment that outliers in the time series have a remarkable negative impact on the online modeling of the time-varying and nonlinear industrial processes. During the learning stage, the support vector clustering (SVC) approach was adopted to detect and remove the outliers and then to obtain the valid data area. Moreover, an improved recursive learning strategy, with nodes online growing and pruning, was developed to enhance the model generalization with a similar computation load. Through an example, the proposed method which was applied to predict the silicon mass fraction in hot metal of blast furnace showed a good performance, with 81% percentage of target hitting and 0.054 7 of the root mean square error of prediction when the size of predicted sample set was 566. These criteria are better than alternative methods, implying that the R-RLSSVR based modeling method is more robust and precise.

Published: 17 May 2012
CLC:  TP 301.6  
  TQ 02  
Cite this article:

YU Hai-qing, LIU Yi, CHEN Kun, JI Jun1, LI Ping. Robust recursive kernel learning modeling method with
application to blast furnace. J4, 2012, 46(4): 705-711.

URL:     OR


针对现场采集时间序列数据中的离群点显著影响时变非线性工业过程在线模型性能这一问题,提出鲁棒的递推最小二乘支持向量机软测量建模方法.在模型训练阶段,采用支持向量聚类(SVC)排除离群点,建立有效的数据区域.将SVC用于递推过程前向学习阶段,并引入更有效的增删节点准则,在快速递推的前提下提高了模型的推广能力.将该方法应用于工业高炉过程铁水的硅质量分数预测,通过试验连续预测566炉高炉铁水硅质量分数,命中率高达81%,预测均方根误差为0.054 7,表明了较其他方法有更好的鲁棒性与精度.

[1] KADLEC P, GABRYS B, STRANDT S. Datadriven soft sensors in the process industry [J]. Computers and Chemical Engineering, 2009, 33(4): 795-814.
[2] KANOA M, NAKAGAWA Y. Databased process monitoring, process control, and quality improvement: recent developments and applications in steel industry [J]. Computers and Chemical Engineering, 2008, 32(1/2): 12-24.
[3] 王华秋,廖晓峰,邹航,等.自反馈RBF网络在高炉热状态模型预测中的应用[J].系统工程与电子技术,2008,30(5): 929-934.
WANG Huaqiu, LIAO Xiaofeng, ZOU Hang, et al. Application of selffeedback RBF NN in prediction model for heat state of blast furnace [J]. Systems Engineering and Electronics, 2008, 30(5): 929-934.
[4] 刘学艺,刘祥官,王文慧.贝叶斯网络在高炉铁水硅含量预测中的应用[J].钢铁,2005,40(3): 17-20.
LIU Xueyi, LIU Xiangguan, WANG Wenhui. Application of bayesian network to predicting silicon content in hot metal [J]. Iron and Steel, 2005, 40(3): 17-20.
[5] 渐令,龚淑华,王义康.基于支持向量机的高炉铁水硅含量多类别分类[J].浙江大学学报:理学版,2007,34(3): 282-285.
JIAN Ling, GONG Shuhua, WANG Yikang. Classifier of silicon content in hot metal based on support vector machines [J]. Journal of Zhejiang University: Science Edition, 2007, 34(3): 282-285.
[6] 赵敏.高炉冶炼过程的复杂性机理及其预测研究[D].杭州: 浙江大学, 2008: 13-32.
ZHAO Min. Complexity mechanism and predictive research for BF ironmaking process [D]. Hangzhou: Zhejiang University, 2008: 13-32.
[7] GAO Chuanhou, CHEN Jiming, ZENG Jiusun, et al. A chaosbased iterated multistep predictor for blast furnace ironmaking process [J]. American Institute of Chemical Engineers, 2009, 55(4): 947-962.
[8] SUYKENS J, VAN GESTEL T, DE BRABANTER J, et al. Least squares support vector machines [M]. Singapore: World Scientific, 2002: 69-103.
[9] TAYLOR J, CRISTIANINI N. Kernel methods for pattern analysis [M]. Cambridge: Cambridge University Press, 2004: 38-75.
[10] WANG Haiqing, LI Ping, GAO Furong, et al. Kernel classifier with adaptive structure and fixed memory for process diagnosis [J]. American Institute of Chemical Engineers Journal, 2006, 52(10): 3515-3531.
[11] 刘毅,陈坤,王海清,等.选择性递推LSSVR及其在过程建模中的应用[J].高校化学工程学报,2008, 22(6): 1043-1048.
LIU Yi, CHEN Kun, WANG Haiqing, et al. Selective recursive LSSVR with applications to process modeling [J]. Journal of Chemical Engineering of Chinese Universities, 2008, 22(6): 1043-1048.
[12] CUI Wentong, YAN Xuefeng. Adaptive weighted least square support vector machine regression integrated with outlier detection and its application in QSAR [J]. Chemometrics and Intelligent Laboratory Systems, 2009, 98(2): 130-135.
[13] WANG Jeenshing, CHIANG Jenchieh. A cluster validity measure with outlier detection for support vector clustering [J]. IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics, 2008, 38(1): 78-89.
[14] GOLUB G, VAN LOAN C. Matrix computations [M]. 3rd ed. Baltimore: The John Hopkins University Press, 1996: 133-186.

[1] LIU Jia-hai, YANG Mao-lin, LEI Hang, LIAO Yong. Multicore real-time task allocation algorithms with shared resource constraints[J]. J4, 2014, 48(1): 113-117.
[2] ZHAO Shi-kui, FANG Shui-liang, GU Xin-jian. Genetic algorithm with new initialization mechanism for flexible job shop scheduling[J]. J4, 2013, 47(6): 1022-1030.
[3] ZHANG Jun-chao, YUE Mao-xiong, LIU Hua-feng. Dynamic PET image reconstruction with Geometrical structure
prior constraints
[J]. J4, 2012, 46(6): 961-966.
[4] LIU Yi, LI Ping, GAO Zeng-liang. Quality prediction of hot metal in blast furnace using improved
support vector regression
[J]. J4, 2012, 46(5): 830-836.
[5] FANG Shui-liang, YAO Yan-fei, ZHAO Shi-kui. Improved genetic algorithm for flexible job shop scheduling[J]. J4, 2012, 46(4): 629-635.
[6] LIU Jia-hai, YANG Mao-lin. Fair scheduling algorithm on multi-core platforms based platforms[J]. J4, 2011, 45(9): 1566-1570.
[7] NI He, Cheng-Gang, SUN Feng-Rui. Adaptive hybrid evolutionary modeling method and its application[J]. J4, 2010, 44(8): 1490-1495.