Home > Research > Publications & Outputs > Two projection methods for use in the analysis ...

Links

Text available via DOI:

View graph of relations

Two projection methods for use in the analysis of multivariate process data with an illustration in petrochemical production

Research output: Contribution to journalJournal articlepeer-review

Published

Standard

Two projection methods for use in the analysis of multivariate process data with an illustration in petrochemical production. / Badcock, J.; Bailey, T.C.; Jonathan, P.; Krzanowski, W.J.

In: Technometrics, Vol. 46, No. 4, 2004, p. 392-403.

Research output: Contribution to journalJournal articlepeer-review

Harvard

APA

Vancouver

Author

Badcock, J. ; Bailey, T.C. ; Jonathan, P. ; Krzanowski, W.J. / Two projection methods for use in the analysis of multivariate process data with an illustration in petrochemical production. In: Technometrics. 2004 ; Vol. 46, No. 4. pp. 392-403.

Bibtex

@article{b9287b48065c4f418e4146aefb9190b0,
title = "Two projection methods for use in the analysis of multivariate process data with an illustration in petrochemical production",
abstract = "Principal components analysis (PCA) is often used in the analysis of multivariate process data to identify important combinations of the original variables on which to focus for more detailed study. However, PCA and other related projection techniques from the standard multivariate repertoire are not explicitly designed to address or to exploit the strong autocorrelation and temporal cross-correlation structures that are often present in multivariate process data. Here we propose two alternative projection techniques that do focus on the temporal structure in such data and that therefore produce components that may have some analytical advantages over those resulting from more conventional multivariate methods. As in PCA, both of our suggested methods linearly transform the original p-variate time series into uncorrelated components; however, unlike PCA, they concentrate on deriving components with particular temporal correlation properties, rather than those with maximal variance. The first technique finds components that exhibit distinctly different autocorrelation structures via modification of a signal-noise decomposition method used in image analysis. The second method draws on ideas from common PCA to produce components that are not only uncorrelated as in PCA, but that also have approximately zero temporally lagged cross-correlations for all time lags. We present the technical details for these two methods, assess their performance through simulation studies, and illustrate their use on multivariate output measures from a fluidized catalytic cracking unit used in petrochemical production, contrasting the results obtained with those from standard PCA.",
keywords = "Multivariate process control, Principal components analysis, Temporally structured components, Temporally uncorrelated components, Computer simulation, Correlation methods, Data reduction, Image analysis, Petrochemicals, Principal component analysis, Process control, Signal to noise ratio, Statistical methods, Multivariate control charts, Data processing",
author = "J. Badcock and T.C. Bailey and P. Jonathan and W.J. Krzanowski",
year = "2004",
doi = "10.1198/004017004000000491",
language = "English",
volume = "46",
pages = "392--403",
journal = "Technometrics",
issn = "0040-1706",
publisher = "American Statistical Association",
number = "4",

}

RIS

TY - JOUR

T1 - Two projection methods for use in the analysis of multivariate process data with an illustration in petrochemical production

AU - Badcock, J.

AU - Bailey, T.C.

AU - Jonathan, P.

AU - Krzanowski, W.J.

PY - 2004

Y1 - 2004

N2 - Principal components analysis (PCA) is often used in the analysis of multivariate process data to identify important combinations of the original variables on which to focus for more detailed study. However, PCA and other related projection techniques from the standard multivariate repertoire are not explicitly designed to address or to exploit the strong autocorrelation and temporal cross-correlation structures that are often present in multivariate process data. Here we propose two alternative projection techniques that do focus on the temporal structure in such data and that therefore produce components that may have some analytical advantages over those resulting from more conventional multivariate methods. As in PCA, both of our suggested methods linearly transform the original p-variate time series into uncorrelated components; however, unlike PCA, they concentrate on deriving components with particular temporal correlation properties, rather than those with maximal variance. The first technique finds components that exhibit distinctly different autocorrelation structures via modification of a signal-noise decomposition method used in image analysis. The second method draws on ideas from common PCA to produce components that are not only uncorrelated as in PCA, but that also have approximately zero temporally lagged cross-correlations for all time lags. We present the technical details for these two methods, assess their performance through simulation studies, and illustrate their use on multivariate output measures from a fluidized catalytic cracking unit used in petrochemical production, contrasting the results obtained with those from standard PCA.

AB - Principal components analysis (PCA) is often used in the analysis of multivariate process data to identify important combinations of the original variables on which to focus for more detailed study. However, PCA and other related projection techniques from the standard multivariate repertoire are not explicitly designed to address or to exploit the strong autocorrelation and temporal cross-correlation structures that are often present in multivariate process data. Here we propose two alternative projection techniques that do focus on the temporal structure in such data and that therefore produce components that may have some analytical advantages over those resulting from more conventional multivariate methods. As in PCA, both of our suggested methods linearly transform the original p-variate time series into uncorrelated components; however, unlike PCA, they concentrate on deriving components with particular temporal correlation properties, rather than those with maximal variance. The first technique finds components that exhibit distinctly different autocorrelation structures via modification of a signal-noise decomposition method used in image analysis. The second method draws on ideas from common PCA to produce components that are not only uncorrelated as in PCA, but that also have approximately zero temporally lagged cross-correlations for all time lags. We present the technical details for these two methods, assess their performance through simulation studies, and illustrate their use on multivariate output measures from a fluidized catalytic cracking unit used in petrochemical production, contrasting the results obtained with those from standard PCA.

KW - Multivariate process control

KW - Principal components analysis

KW - Temporally structured components

KW - Temporally uncorrelated components

KW - Computer simulation

KW - Correlation methods

KW - Data reduction

KW - Image analysis

KW - Petrochemicals

KW - Principal component analysis

KW - Process control

KW - Signal to noise ratio

KW - Statistical methods

KW - Multivariate control charts

KW - Data processing

U2 - 10.1198/004017004000000491

DO - 10.1198/004017004000000491

M3 - Journal article

VL - 46

SP - 392

EP - 403

JO - Technometrics

JF - Technometrics

SN - 0040-1706

IS - 4

ER -