Please wait a minute...
浙江大学学报(工学版)
J4  2010, Vol. 44 Issue (4): 652-658    DOI: 10.3785/j.issn.1008-973X.2010.04.005
    
PCA-SVDD based fault detection and selflearning identification
ZHU Zhibo1, WANG Peiliang2, SONG Zhihuan1
1. State Key Laboratory of Industrial Control Technology, Institute of Industrial Process Control, Zhejiang University,
Hangzhou 310027, China; 2. School of Information Engineering, Huzhou Teachers College, Huzhou 313000, China
Download:   PDF(0KB) HTML
Export: BibTeX | EndNote (RIS)      

Abstract  

In order to utilize the advantage of fault detection and to overcome the shortage of fault identification and diagnosis in the multivariable statistical process control, a new algorithm for chemical process fault detection and selflearning identification was proposed. Principal component analysis (PCA) was applied for fault detection and feature extraction, three pattern discriminant methods based on principal component analysissupport vector data description (PCASVDD) for fault selflearning identification were proposed. Considering the faults distribution overlapping problem which happened possibly in fault identification, two methods based on Euclid distance and unitary radius were analyzed and compared. A method based on weighted unitary radius for novel fault identification was also proposed. A case study of Tennessee Eastman (TE) process showed the feasibility and effectiveness of the proposed fault detection and selflearning identification algorithm.

Published: 14 May 2010
CLC:  TP277  
Service
E-mail this article
Add to my bookshelf
Add to citation manager
E-mail Alert
RSS
Articles by authors
Cite this article:

CHU Zhi-Bo, WANG Pei-Liang, SONG Zhi-Huan. PCA-SVDD based fault detection and selflearning identification. J4, 2010, 44(4): 652-658.

URL:

http://www.zjujournals.com/eng/10.3785/j.issn.1008-973X.2010.04.005     OR     http://www.zjujournals.com/eng/Y2010/V44/I4/652


基于PCASVDD的故障检测和自学习辨识

为了利用多变量统计过程控制在故障检测上的优势以及克服其在故障辨识诊断上的缺陷,提出一套新的用于化工过程的故障检测和自学习辨识算法.应用主元分析（PCA）实施故障检测并对故障数据运用PCA特征提取,提出3种基于主元分析支持向量数据描述（PCASVDD）的模式判别方法来实现故障的自学习辨识：考虑故障辨识时可能出现的类分布重合问题,分析和比较了基于欧氏距离和归一化半径判别这2种方法,提出针对新型未知故障辨识的加权归一化半径判别法.通过对Tennessee Eastman（TE）过程的仿真研究,说明了提出的故障检测和自学习辨识算法的可行性和有效性.

［1］ 张杰,阳宪惠.多变量统计过程控制［M］.北京：化学工业出版社,2000: 2427.
［2］ CHIANG L H, RUSSELL E L, BRAATZ R D. Fault detection and diagnosis in industrial systems ［M］. London: SpringleVerlag, 2001: 1598.
［3］ 陈国金,梁军,刘育明,等.基于多元统计投影方法的过程监控技术研究［J］.浙江大学学报：工学版,2004,38(12): 15611565.
CHEN Guojin, LIANG Jun, LIU Yuming, et al. Process monitoring based on multivariate statistical projection analysis ［J］. Journal of Zhejiang University: Engineering Science, 2004, 38(12): 15611565.
［4］ 葛志强,宋执环,杨春节.基于MCUSUMICAPCA的微小故障检测［J］.浙江大学学报：工学版,2008, 42(3): 373377.
GE Zhiqiang, SONG Zhihuan, YANG Chunjie. Small shift detection based on MCUSUMICAPCA ［J］. Journal of Zhejiang University: Engineering Science, 2008, 42(3): 373377.
［5］ DUNIA R, QIN S J. Subspace approach to multidimensional fault identification and reconstruction ［J］. AIChE Journal, 1998, 44(8): 18131831.
［6］ MAURYAA M R, RENGASWAMYB R, VENKATASUBRAMANIANA V. A signed directed graphbased systematic framework for steadystate malfunction diagnosis inside control loops ［J］. Chemical Engineering Science, 2006, 61(6): 17901810.
［7］ TAX M J D, DUIN P W R. Support vector domain description ［J］. Pattern Recognition Letters, 1999, 20(1113): 11911199.
［8］ TAX M J D, DUIN P W R. Support vector data description ［J］. Machine Learning, 2004, 54(1): 4566.
［9］ 潘明清,周晓军,吴瑞明,等.基于主元分析的支持向量数据描述机械故障诊断［J］.传感技术学报,2006, 19(1): 128131.
PAN Mingqing, ZHOU Xiaojun, WU Ruiming, et al. Research of machine fault diagnosis based on PCA and SVDD ［J］. Chinese Journal of Sensors and Actuators, 2006, 19(1): 128131.
［10］ 张庆,徐光华,王晶,等.基于支持向量域描述的多故障诊断动态模型［J］.西安交通大学学报,2007, 41(5): 594597.
ZHANG Qing, XU Guanghua, WANG jing, et al. Dynamic multifault diagnosis model based on support vector domain description ［J］. Journal of Xian Jiaotong University, 2007, 41(5): 594597.
［11］ DOWNS J J, VOGEL E F. A plantwide industrial process control problem ［J］. Computers and Chemical Engineering, 1993, 17(3): 245255.
［12］ Richard Braatz, the Braatz Research Group. TE process simulator software. ［EB/OL］. ［20020422］. http:∥ brahms.scs.uiuc.edu/lssrl/software/TE_process.
[1] LIN Yong, ZHOU Xiao-Jun, ZHANG Wen-Bin, YANG Xian-Yong. Rolling bearing fault diagnosis based on morphological
wavelet theory and bispectrum analysis[J]. J4, 2010, 44(3): 432-439.