Monitoring Nonstationary Processes Using Stationary Subspace Analysis and Fractional Integration Order Estimation

Yuanling Lin, Uwe Kruger, Fengshou Gu, Andrew Ball, Qian Chen

Research output: Contribution to journalArticle

Abstract

This article introduces a framework to monitor complex dynamic and mildly nonstationary processes that are driven by a set of latent factors that can have different integration orders. The framework (i) relies on a novel deflation-based stationary subspace analysis that extracts latent source variables from recorded data sets in an iterative manner and (ii) utilizes the exact local Whittle estimator to calculate the fractional integration orders of the extracted source variables. The framework is embedded within a multivariate time-series structure to model the dynamic characteristics of the latent factors and to remove serial correlation in order to construct univariate monitoring statistics. A numerical and an industrial case study show that this framework is capable of modeling dynamic and mildly nonstationary variable inter-relationships that can have different integration orders.

LanguageEnglish
Pages6486-6504
Number of pages19
JournalIndustrial and Engineering Chemistry Research
Volume58
Issue number16
Early online date19 Mar 2019
DOIs
Publication statusPublished - 24 Apr 2019

Fingerprint

Process monitoring
Time series
Statistics
Monitoring

Cite this

@article{566bb3673be84e00ab52d2160ec57ea2,
title = "Monitoring Nonstationary Processes Using Stationary Subspace Analysis and Fractional Integration Order Estimation",
abstract = "This article introduces a framework to monitor complex dynamic and mildly nonstationary processes that are driven by a set of latent factors that can have different integration orders. The framework (i) relies on a novel deflation-based stationary subspace analysis that extracts latent source variables from recorded data sets in an iterative manner and (ii) utilizes the exact local Whittle estimator to calculate the fractional integration orders of the extracted source variables. The framework is embedded within a multivariate time-series structure to model the dynamic characteristics of the latent factors and to remove serial correlation in order to construct univariate monitoring statistics. A numerical and an industrial case study show that this framework is capable of modeling dynamic and mildly nonstationary variable inter-relationships that can have different integration orders.",
author = "Yuanling Lin and Uwe Kruger and Fengshou Gu and Andrew Ball and Qian Chen",
year = "2019",
month = "4",
day = "24",
doi = "10.1021/acs.iecr.8b05099",
language = "English",
volume = "58",
pages = "6486--6504",
journal = "Industrial & Engineering Chemistry Research",
issn = "0888-5885",
publisher = "American Chemical Society",
number = "16",

}

Monitoring Nonstationary Processes Using Stationary Subspace Analysis and Fractional Integration Order Estimation. / Lin, Yuanling; Kruger, Uwe; Gu, Fengshou; Ball, Andrew; Chen, Qian.

In: Industrial and Engineering Chemistry Research, Vol. 58, No. 16, 24.04.2019, p. 6486-6504.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Monitoring Nonstationary Processes Using Stationary Subspace Analysis and Fractional Integration Order Estimation

AU - Lin, Yuanling

AU - Kruger, Uwe

AU - Gu, Fengshou

AU - Ball, Andrew

AU - Chen, Qian

PY - 2019/4/24

Y1 - 2019/4/24

N2 - This article introduces a framework to monitor complex dynamic and mildly nonstationary processes that are driven by a set of latent factors that can have different integration orders. The framework (i) relies on a novel deflation-based stationary subspace analysis that extracts latent source variables from recorded data sets in an iterative manner and (ii) utilizes the exact local Whittle estimator to calculate the fractional integration orders of the extracted source variables. The framework is embedded within a multivariate time-series structure to model the dynamic characteristics of the latent factors and to remove serial correlation in order to construct univariate monitoring statistics. A numerical and an industrial case study show that this framework is capable of modeling dynamic and mildly nonstationary variable inter-relationships that can have different integration orders.

AB - This article introduces a framework to monitor complex dynamic and mildly nonstationary processes that are driven by a set of latent factors that can have different integration orders. The framework (i) relies on a novel deflation-based stationary subspace analysis that extracts latent source variables from recorded data sets in an iterative manner and (ii) utilizes the exact local Whittle estimator to calculate the fractional integration orders of the extracted source variables. The framework is embedded within a multivariate time-series structure to model the dynamic characteristics of the latent factors and to remove serial correlation in order to construct univariate monitoring statistics. A numerical and an industrial case study show that this framework is capable of modeling dynamic and mildly nonstationary variable inter-relationships that can have different integration orders.

UR - http://www.scopus.com/inward/record.url?scp=85064972190&partnerID=8YFLogxK

U2 - 10.1021/acs.iecr.8b05099

DO - 10.1021/acs.iecr.8b05099

M3 - Article

VL - 58

SP - 6486

EP - 6504

JO - Industrial & Engineering Chemistry Research

T2 - Industrial & Engineering Chemistry Research

JF - Industrial & Engineering Chemistry Research

SN - 0888-5885

IS - 16

ER -