Fault detection and identification in chemical processes based on feature engineering and kernel extreme learning machine

被引：0
|
作者：
Ren Y.-J. [1 ]
Wang J. [1 ]
Tian W.-D. [1 ]
机构：
[1] College of Chemical Engineering, Qingdao University of Science & Technology, Qingdao
来源：
Gao Xiao Hua Xue Gong Cheng Xue Bao/Journal of Chemical Engineering of Chinese Universities | 2019年 / 33卷 / 05期
关键词：
Fault detection and recognition; Genetic algorithm; Independent component analysis; Kernel extreme learning machine; Mutual information;
D O I：
10.3969/j.issn.1003-9015.2019.05.031
中图分类号：
学科分类号：
摘要：
A combination of feature engineering and kernel extreme learning machine (KELM) was proposed to detect and identify chemical process faults that possess features of complexity, non-Gaussian data and strong coupling. Original data was first mapped into independent directions by independent component analysis (ICA) algorithm, and the statistics of the data was calculated to identify the fault. Moreover, in order to improve the accuracy of the classification, every two samples were compared with each other using mutual information (MI), and the core variables were selected by contribution degree and correlation. Finally, the acquired features were input into KELM for fault classification. Genetic algorithm was used to optimize the parameters to improve diagnostic accuracy. Application of this method to the Tennessee Eastman (TE) process and an industrial depropanation process indicates that the proposed method can effectively detect faults and accurately identify fault types. © 2019, Editorial Board of "Journal of Chemical Engineering of Chinese Universities". All right reserved.
引用
收藏
页码：1271 / 1284
页数：13
相关论文
共 29 条
  • [1] Li C.K., Zhao D.F., Wang C.L., Et al., Study on a knowledge automation method for fault diagnosis of chemical processes, Journal of Chemical Engineering of Chinese Universities, 32, 2, pp. 393-400, (2018)
  • [2] Tang G.Z., Zhang G.M., Gong J.M., A new method for fault diagnosis of condenser vacuum based on fuzzy rough set and case-based reasoning, CIESC Journal, 62, 8, pp. 2227-2231, (2011)
  • [3] Zhou S., Han Q.L., Hu Z., Et al., Pattern recognition method for proton exchange membrane fuel cell fault diagnosis, Journal of Tongji University (Natural Science Edition), 45, 3, pp. 104-108, (2017)
  • [4] Huang D., Song X., Application of neural networks to chemical fault diagnosis, Journal of Process Control, 13, 1, pp. 6-9, (2006)
  • [5] Guo J.Y., Wang X., Li Y., Fault detection in chemical processes using weighted differential principal component analysis, Journal of Chemical Engineering of Chinese Universities, 32, 1, pp. 183-192, (2018)
  • [6] Xu Y., Deng X.G., Zhong N., A fault diagnosis method for multimode processes based on ICA mixture models, CIESC Journal, 67, 9, pp. 3793-3803, (2016)
  • [7] Lyu Y., Deng Y.J., Yang H.Z., Compositional support vector machine model based on feature extraction of categories, CIESC Journal, 62, 8, pp. 2164-2169, (2011)
  • [8] Wu Y.F., Wang Z.L., Industrial process monitoring and fault diagnosis with nonlinear independent component analysis based on kurtosis, Journal of East China University of Science and Technology, 34, 4, pp. 568-573, (2008)
  • [9] Long Y., He Y.G., Zhang Z., Et al., Switched Current Circuit Fault Diagnosis Based on Wavelet Transform and ICA Feature Extraction, Journal of Instrumentation, 36, 10, pp. 2389-2400, (2015)
  • [10] Zhang Y., Yang Z., Fault detection of non-Gaussian processes based on modified independent component analysis, Chemical Engineering Science, 65, 16, pp. 4630-4639, (2010)
123