Abstract
Conventional kernel principal component analysis (KPCA) does not always perform well for nonlinear process monitoring because the beneficial information for fault detection may be submerged under the retained kernel principal components (KPCs). To overcome this deficiency, an adaptively weighted KPCA integrated with support vector data description (WKPCA-SVDD) monitoring method is proposed. In WKPCA-SVDD, the importance of each KPC is evaluated online by the change rate of T2 statistic and then distinguished weighting values are set on the KPCs. The behaviors of all KPCs are comprehensively evaluated by the SVDD technique. Since the beneficial information is highlighted, the monitoring performance of the statistic in the dominant subspace can be improved. The proposed WKPCA-SVDD is applied to both a numerical process and the complicated Tennessee Eastman benchmark process. Monitoring results have indicated the efficiency of the WKPCA-SVDD method.
Similar content being viewed by others
References
D. Aguado, A. Ferrer, J. Ferrer and A. Seco, Chemom. Intell. Lab. Syst., 85, 82 (2007).
A. AlGhazzawi and B. Lennox, Control Eng. Pract., 16, 294 (2008).
L. H. Chiang, E. L. Russell and R. D. Braatz, Chemom. Intell. Lab. Syst., 50, 243 (2000).
J.-M. Lee, C. Yoo, S.W. Choi, P. A. Vanrolleghem and I.-B. Lee, Chem. Eng. Sci., 59, 223 (2004).
J.-M. Lee, C. Yoo and I.-B. Lee, J. Process Control, 14, 467 (2004).
P. Odiowei and Y. Cao, Chemom. Intell. Lab. Syst., 103, 59 (2010).
E. Tomba, P. Facco, F. Bezzo, S. García-Muñoz and M. Barolo, Chemom. Intell. Lab. Syst., 116, 67 (2012).
K. Han, K. J. Park, H. Chae and E. S. Yoon, Korean J. Chem. Eng., 25, 13 (2008).
M. H. Kim and C. K. Yoo, Korean J. Chem. Eng., 25, 947 (2008).
Q. Jiang and X. Yan, AIChE J., 60, 949 (2014).
L.H. Chiang, E. Russell and R. D. Braatz, Fault detection and diagnosis in industrial systems, Springer Verlag, London (2001).
Q. Jiang and X. Yan, Chemom. Intell. Lab. Syst., 119, 11 (2012).
J.V. Kresta, J. F. Macgregor and T. E. Marlin, Can. J. Chem. Eng., 69, 35 (2009).
Q. Jiang, X. Yan and W. Zhao, Ind. Eng. Chem. Res., 52, 1635 (2013).
Q. Chen, U. Kruger, M. Meronk and A. Leung, Control Eng. Pract., 12, 745 (2004).
I. T. Jolliffe, Applied Statistics, 300 (1982).
H. Wang, Z. Song and P. Li, Ind. Eng. Chem. Res., 41, 2455 (2002).
S. Wold, Chemom. Intell. Lab. Syst., 23, 149 (1994).
X. B. He, Y. P. Yang and Y. H. Yang, Chemom. Intell. Lab. Syst., 93, 27 (2008).
D. L. Ferreira, S. Kittiwachana, L.A. Fido, D. R. Thompson, R. E. Escott and R. G. Brereton, Analyst, 134, 1571 (2009).
Q. Jiang and X. Yan, Chemom. Intell. Lab. Syst., 127, 121 (2013).
C. Alzate and J. A. Suykens, Pattern Analysis and Machine Intelligence, IEEE Transactions on, 32, 335 (2010).
Q. Jiang, X. Yan, Z. Lv and M. Guo, Korean J. Chem. Eng., 30, 1181 (2013).
M. A. Kramer, AIChE J., 37, 233 (1991).
D. Dong and T. J. McAvoy, Comput. Chem. Eng., 20, 65 (1996).
H. Hiden, M. Willis, M. Tham and G. Montague, Comput. Chem. Eng., 23, 413 (1999).
B. Schölkopf, A. Smola and K.-R. Müller, Neural Computation, 10, 1299 (1998).
P. Cui, J. Li and G. Wang, Expert Systems with Applications, 34, 1210 (2008).
Z. Ge and Z. Song, Control Eng. Pract., 16, 1427 (2008).
Z. Ge, C. Yang and Z. Song, Chem. Eng. Sci., 64, 2245 (2009).
V. H. Nguyen and J.-C. Golinval, Eng. Struct., 32, 3683 (2010).
Z. Ge and Z. Song, Expert Syst. Appl., 38, 9821 (2011).
Y. Yang and J. M. Lee, J. Process Control, 23, 852 (2013).
Z. Ge, F. Gao and Z. Song, J. Process Control, 21, 949 (2011).
D.M. Tax and R. P. Duin, Machine Learning, 54, 45 (2004).
J. J. Downs and E. F. Vogel, Comput. Chem. Eng., 17, 245 (1993).
P. R. Lyman and C. Georgakis, Comput. Chem. Eng., 19, 321 (1995).
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Jiang, Q., Yan, X. Improved fault detection in nonlinear chemical processes using WKPCA-SVDD. Korean J. Chem. Eng. 31, 1935–1942 (2014). https://doi.org/10.1007/s11814-014-0151-y
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11814-014-0151-y