Adaptive soft sensors using local partial least squares with moving window approach

Developing soft sensors for an industrial process, the colinearity of the predictor variables and the time-varying nature of the process need to be concerned. In many industrial applications, the partial least square (PLS) has been proven to be able to capture the linear relationship between input and output variables for a local operating region; therefore, the PLS model needs to be adapted to accommodate the time-varying nature of the process. In this paper, a fast-moving window algorithm is derived to update the PLS model. The proposed approach adapted the parameters of the inferential model with the dissimilarities between the new and oldest data and incorporated them into the kernel algorithm for the PLS. The computational loading of the model adaptation was therefore independent of the window size. Because a moving window approach is sensitive to outliers, the confidence intervals for the primary variables were created on the basis of the prediction uncertainty. The inferential model would not be misled by the outliers from the online analyzers, whereas the model could be updated during the transition stage. An industrial example, predicting oxygen concentrations in the air separation process, demonstrated the effectiveness of the proposed approach for the process industry. Copyright © 2012 Curtin University of Technology and John Wiley & Sons, Ltd.

[1]  Jie Zhang,et al.  Nonlinear projection to latent structures method and its applications , 2006 .

[2]  A. Höskuldsson.  PLS regression methods , 1988 .

[3]  C. Yoo,et al.  Nonlinear PLS modeling with fuzzy inference system , 2002 .

[4]  Bogdan Gabrys,et al.  Data-driven Soft Sensors in the process industry , 2009, Comput. Chem. Eng..

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

[6]  Bhupinder S. Dayal,et al.  Recursive exponentially weighted PLS and its applications to adaptive control and prediction , 1997 .

[7]  Xi Zhang,et al.  Nonlinear Multivariate Quality Estimation and Prediction Based on Kernel Partial Least Squares , 2008 .

[8]  S. Qin.  Recursive PLS algorithms for adaptive data modeling , 1998 .

[9]  Edward C. Malthouse,et al.  Nonlinear partial least squares , 1997 .

[10]  Alison J. Burnham,et al.  Frameworks for latent variable multivariate regression , 1996 .

[11]  Thomas J. McAvoy,et al.  Nonlinear FIR Modeling via a Neural Net PLS Approach , 1996 .

[12]  C. Braak,et al.  Comments on the PLS kernel algorithm , 1994 .

[13]  S. Wold,et al.  The kernel algorithm for PLS , 1993 .

[14]  Ping Wu,et al.  Online dual updating with recursive PLS model and its application in predicting crystal size of purified terephthalic acid (PTA) process , 2006 .

[15]  S. Wold,et al.  PLS-regression: a basic tool of chemometrics , 2001 .

[16]  Xiao Fan Wang,et al.  Soft sensing modeling based on support vector machine and Bayesian model selection , 2004, Comput. Chem. Eng..

[17]  Bogdan Gabrys,et al.  Review of adaptation mechanisms for data-driven soft sensors , 2011, Comput. Chem. Eng..

[18]  Ahmet Palazoglu,et al.  Orthogonal nonlinear partial least-squares regression , 2003 .

[19]  R. Gudi,et al.  Identification on Demand Using a Blockwise Recursive Partial Least-Squares Technique† , 2003 .

[20]  S. Wold.  Nonlinear partial least squares modelling II. Spline inner relation , 1992 .

[21]  E. Martin,et al.  Non-linear projection to latent structures revisited: the quadratic PLS algorithm , 1999 .

[22]  A. J. Morris,et al.  Non-linear projection to latent structures revisited (the neural network PLS algorithm) , 1999 .

[23]  Bhupinder S. Dayal,et al.  Improved PLS algorithms , 1997 .

[24]  S. Wold,et al.  Nonlinear PLS modeling , 1989 .

[25]  Bruce R. Kowalski,et al.  PREDICTION ERROR IN LEAST SQUARES REGRESSION : FURTHER CRITIQUE ON THE DEVIATION USED IN THE UNSCRAMBLER , 1996 .

[26]  Rubens Maciel Filho,et al.  Soft sensors development for on-line bioreactor state estimation , 2000 .

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

[28]  Salvador García-Muñoz,et al.  A comparison of different methods to estimate prediction uncertainty using Partial Least Squares (PLS): A practitioner's perspective , 2009 .