A New Method of On-Line Fault Monitoring Based on Nonlinear MPCA

Abstract:

Multiway principal components analysis (MPCA) is a linear model in nature, thus, limited when it is applied to batch process. In this paper, the linear model MPCA was complemented with an autoassociative neural network model in order to generate nonlinear principal components. The networks bottleneck layer outputs (nonlinear principal components) were made orthogonal. A method to estimate confidence limits based on a kernel probability density function was proposed since the nonlinear scores are no normally distributed. A statistic-like parameter () was proposed to evaluate on-line scores for new runs using the density estimated confidence bounds and replacing the statistic. The proposed method was applied to monitoring fed-batch streptomycete production, and the simulation results show that the nonlinear scores obtained with the autoassociative neural networks capture more process data variance than if obtained with a linear method and the density estimation method proved to be more reliable.

• Periodical: Advanced Materials Research (Volume 486)
• Pages: 472-477
• DOI: https://doi.org/10.4028/www.scientific.net/AMR.486.472
• Online since: March 2012
• Authors: Ying Wang Xiao, Yang Jun
• Keywords: Batch Process, Fault Monitoring, Kernel Density Estimation, MPCA, Neural Network (NN)

References

[1] Nomikos P, MacGregor J F.: Monitoring batch process using multi-way principle component analysis, Journal of America Institute Chemical Engineer, Vol. 40(8) (1994), pp.1361-1375.

[2] He Ning, Wang Shuqing, Xie Lei: An improved adaptive MPCA for monitoring streptomycin fermentation process, Journal of Chemical Industry and Engineering (China), Vol. 12(1) (2004), pp.96-101.

[3] Jong-Min Lee, Chang Kyoo Yoo, In-Beum Lee: Enhanced process monitoring of fed-batch penicillin cultivation using time-varying and multivariate statistical analysis, Journal of Biotechnology, Vol. 110 (2004), pp.119-136.

DOI: 10.1016/j.jbiotec.2004.01.016

[4] M. Daszykowski, B. Walczak, D. L. Massart : A journey into low-dimensional spaces with autoassociative neural networks, Talanta, Vol. 59 (2008), pp.1095-1105.

DOI: 10.1016/s0039-9140(03)00018-3

[5] C. Aldrich: Visualization of transformed multivariate data sets with autoassociative neural networks, Pattern Recognition, Vol. 19 (1998), pp.749-764.

DOI: 10.1016/s0167-8655(98)00054-3

[6] Fuat Doymaz, James Chen, Jose A. Romagnoli, Ahmet Palazoglu: A robust strategy for real-time process monitoring, Journal of Process Control, Vol. 11 (2001), pp.343-359.

DOI: 10.1016/s0959-1524(00)00004-4

[7] Q. chen, R. J. Wynne, P. Goulding, D. Sandoz: The application of principal component analysis and kernel density estimation to enhance process monitoring, Control Engineering Practice, Vol. 8 (2000), pp.531-543.

DOI: 10.1016/s0967-0661(99)00191-4

[8] Q. Chen, R. Wynne, D. J. Sandoz, & U. Kruger: Improved Kernel density estimation for clustered data using additive noise, Proceedings of the American control conference, Chicago, IL, USA, (2010), pp.28-30.

DOI: 10.1109/acc.2000.876733

[9] A. W. Bowman: Alternative method of cross-validation for the smoothing of density estimate, Biometrika, Vol. 76 (2009), p.353.

[10] J. Foley, A. Dam, S. Feiner, J. Hughes, In: C. Reading (Eds. ): Computer Graphics: Principles and Practice, Addison-Wesley Publishing Company, New York (2000), p.1175.