Dynamic processes monitoring using recursive kernel principal component analysis

Abstract The dynamic process monitoring is discussed in this paper. Kernel principal component analysis (KPCA) is a nonlinear monitoring method that cannot be employed for dynamic systems. Recursive KPCA (RKPCA) is proposed to monitor the dynamic processes, which is adaptive monitoring method by computing recursively the eigenvalues and eigenvectors in the kernel space when the training data are updated dynamically. The contributions of this article are as follows: (1) The model of history data is used to build new model after the new sample is obtained. The expensive computation is avoided in this article. (2) New nonlinear modeling method is proposed based on a new singular value decomposition (SVD) technique. The results are interesting due to the nonlinear time evolution of the variables involved. The proposed algorithm was applied to the continuous annealing process and penicillin fermentation process for adaptive monitoring and RKPCA could efficiently capture the time-varying and nonlinear relationship in process variables.

[1]  R. Cattell.  The Scree Test For The Number Of Factors. , 1966, Multivariate behavioral research.

[2]  Shufeng Tan,et al.  Reducing data dimensionality through optimizing neural network inputs , 1995 .

[3]  Chun-Chin Hsu,et al.  Adaptive Kernel Principal Component Analysis (KPCA) for Monitoring Small Disturbances of Nonlinear Processes , 2010 .

[4]  S. Joe Qin,et al.  Fault Detection of Nonlinear Processes Using Multiway Kernel Independent Component Analysis , 2007 .

[5]  John F. MacGregor,et al.  Multivariate SPC charts for monitoring batch processes , 1995 .

[6]  Elaine B. Martin,et al.  Non-linear principal components analysis with application to process fault detection , 2000, Int. J. Syst. Sci..

[7]  Mauricio Maestri,et al.  Kernel PCA Performance in Processes with Multiple Operation Modes , 2009 .

[8]  Zhiqiang Ge,et al.  Multivariate Statistical Process Control , 2012 .

[9]  Tordis E. Morud,et al.  Multivariate statistical process control; example from the chemical process industry , 1996 .

[10]  Jie Zhang,et al.  Performance monitoring of processes with multiple operating modes through multiple PLS models , 2006 .

[11]  Edmund R. Malinowski,et al.  Factor Analysis in Chemistry , 1980 .

[12]  Hsiao-Ping Huang,et al.  Fault detection and isolation for dynamic processes using recursive principal component analysis (PCA) based on filtering of signals , 2007 .

[13]  Ralph R. Martin,et al.  Merging and Splitting Eigenspace Models , 2000, IEEE Trans. Pattern Anal. Mach. Intell..

[14]  Manabu Kano,et al.  Monitoring independent components for fault detection , 2003 .

[15]  Hong Zhou,et al.  Decentralized Fault Diagnosis of Large-Scale Processes Using Multiblock Kernel Partial Least Squares , 2010, IEEE Transactions on Industrial Informatics.

[16]  Nola D. Tracy,et al.  Multivariate Control Charts for Individual Observations , 1992 .

[17]  Barry Lennox,et al.  Monitoring a complex refining process using multivariate statistics , 2008 .

[18]  D. W. Osten,et al.  Selection of optimal regression models via cross‐validation , 1988 .

[19]  Jérôme Harmand,et al.  Minimal time bioremediation of natural water resources , 2010, Autom..

[20]  H. Akaike.  A new look at the statistical model identification , 1974 .

[21]  G. Irwin,et al.  Process monitoring approach using fast moving window PCA , 2005 .

[22]  Thomas Voegtlin,et al.  Recursive principal components analysis , 2005, Neural Networks.

[23]  Bernhard Schölkopf,et al.  Nonlinear Component Analysis as a Kernel Eigenvalue Problem , 1998, Neural Computation.

[24]  G. A. Fodor,et al.  From the Editor-in-Chief Industrial Informatics: Predicting with Abstractions , 2005 .

[25]  Zhi-huan Song,et al.  Process Monitoring Based on Independent Component Analysis - Principal Component Analysis ( ICA - PCA ) and Similarity Factors , 2007 .

[26]  T. McAvoy,et al.  Nonlinear principal component analysis—Based on principal curves and neural networks , 1996 .

[27]  In-Beum Lee,et al.  Fault detection and diagnosis based on modified independent component analysis , 2006 .

[28]  Uwe Kruger,et al.  Recursive partial least squares algorithms for monitoring complex industrial processes , 2003 .

[29]  Lamiaa M. Elshenawy,et al.  Efficient Recursive Principal Component Analysis Algorithms for Process Monitoring , 2010 .

[30]  A. Johnson,et al.  The control of fed-batch fermentation processes - A survey , 1987, Autom..

[31]  Zhiqiang Ge,et al.  Improved kernel PCA-based monitoring approach for nonlinear processes , 2009 .

[32]  Ying-wei Zhang,et al.  Combining Kernel Partial Least-Squares Modeling and Iterative Learning Control for the Batch-to-Batch Optimization of Constrained Nonlinear Processes , 2010 .

[33]  David J. Sandoz,et al.  Extended PLS approach for enhanced condition monitoring of industrial processes , 2001 .

[34]  U. Kruger,et al.  Moving window kernel PCA for adaptive monitoring of nonlinear processes , 2009 .

[35]  S. Qin,et al.  Improved nonlinear fault detection technique and statistical analysis , 2008 .

[36]  Richard I. Shrager,et al.  Titration of individual components in a mixture with resolution of difference spectra, pKs, and redox transitions , 1982 .

[37]  S. Wold.  Cross-Validatory Estimation of the Number of Components in Factor and Principal Components Models , 1978 .

[38]  M. Kramer.  Nonlinear principal component analysis using autoassociative neural networks , 1991 .

[39]  Richard D. Braatz,et al.  Fault Detection and Diagnosis in Industrial Systems , 2001 .

[40]  Barry M. Wise,et al.  Development and Benchmarking of Multivariate Statistical Process Control Tools for a Semiconductor Etch Process: Improving Robustness through Model Updating , 1997 .

[41]  Andrei Zinovyev,et al.  Principal Manifolds for Data Visualization and Dimension Reduction , 2007 .

[42]  Chi Ma,et al.  Fault diagnosis of nonlinear processes using multiscale KPCA and multiscale KPLS , 2011 .

[43]  T. Kailath,et al.  Fast Estimation of Principal Eigenspace Using LanczosAlgorithm , 1994 .

[44]  Gunnar Rätsch,et al.  Input space versus feature space in kernel-based methods , 1999, IEEE Trans. Neural Networks.