Fault diagnosis using concurrent projection to latent structures

Yu C. PAN; Yining DONG; S. Joe QIN

doi:10.1016/j.ifacol.2015.09.144

Fault diagnosis using concurrent projection to latent structures

Yu C. PAN, Yining DONG, S. Joe QIN

Research output: Journal Publications › Journal Article (refereed) › peer-review

3 Citations (Scopus)

Abstract

Recently, a concurrent projection to latent structures (CPLS) for multivariate statistical process was proposed. It has been proved to be a better monitoring method than the traditional PLS. However, its fault diagnosis methods have not been developed yet. In this paper, we discuss a new fault diagnosis approach based on CPLS. Five monitoring indices used in CPLS are unified into two general forms. Based on these general forms, we define their complete decomposition contributions (CDC) and reconstruction-based contributions (RBC). The diagnosability of these two contribution methods is further analyzed. Finally, simulation case studies are presented to demonstrate the results.

Original language	English
Pages (from-to)	1276-1281
Number of pages	6
Journal	IFAC-PapersOnLine
Volume	48
Issue number	8
Early online date	25 Sept 2015
DOIs	https://doi.org/10.1016/j.ifacol.2015.09.144
Publication status	Published - 2015
Externally published	Yes
Event	9th IFAC Symposium on Advanced Control of Chemical Processes, ADCHEM 2015 - , Canada Duration: 7 Jun 2015 → 10 Jun 2015

Keywords

Concurrent projection to latent structures (CPLS)
Contribution plots
Data-driven
Fault diagnosis
Process monitoring
Quality monitoring

Access to Document

10.1016/j.ifacol.2015.09.144Licence: CC BY-NC-ND

Cite this

@article{4dc2f0d0f9e74f0cb63aec373027c052,

title = "Fault diagnosis using concurrent projection to latent structures",

abstract = "Recently, a concurrent projection to latent structures (CPLS) for multivariate statistical process was proposed. It has been proved to be a better monitoring method than the traditional PLS. However, its fault diagnosis methods have not been developed yet. In this paper, we discuss a new fault diagnosis approach based on CPLS. Five monitoring indices used in CPLS are unified into two general forms. Based on these general forms, we define their complete decomposition contributions (CDC) and reconstruction-based contributions (RBC). The diagnosability of these two contribution methods is further analyzed. Finally, simulation case studies are presented to demonstrate the results.",

keywords = "Concurrent projection to latent structures (CPLS), Contribution plots, Data-driven, Fault diagnosis, Process monitoring, Quality monitoring",

author = "PAN, {Yu C.} and Yining DONG and QIN, {S. Joe}",

year = "2015",

doi = "10.1016/j.ifacol.2015.09.144",

language = "English",

volume = "48",

pages = "1276--1281",

journal = "IFAC-PapersOnLine",

issn = "2405-8963",

publisher = "IFAC Secretariat",

number = "8",

note = "9th IFAC Symposium on Advanced Control of Chemical Processes, ADCHEM 2015 ; Conference date: 07-06-2015 Through 10-06-2015",

}

TY - JOUR

T1 - Fault diagnosis using concurrent projection to latent structures

AU - PAN, Yu C.

AU - DONG, Yining

AU - QIN, S. Joe

PY - 2015

Y1 - 2015

N2 - Recently, a concurrent projection to latent structures (CPLS) for multivariate statistical process was proposed. It has been proved to be a better monitoring method than the traditional PLS. However, its fault diagnosis methods have not been developed yet. In this paper, we discuss a new fault diagnosis approach based on CPLS. Five monitoring indices used in CPLS are unified into two general forms. Based on these general forms, we define their complete decomposition contributions (CDC) and reconstruction-based contributions (RBC). The diagnosability of these two contribution methods is further analyzed. Finally, simulation case studies are presented to demonstrate the results.

AB - Recently, a concurrent projection to latent structures (CPLS) for multivariate statistical process was proposed. It has been proved to be a better monitoring method than the traditional PLS. However, its fault diagnosis methods have not been developed yet. In this paper, we discuss a new fault diagnosis approach based on CPLS. Five monitoring indices used in CPLS are unified into two general forms. Based on these general forms, we define their complete decomposition contributions (CDC) and reconstruction-based contributions (RBC). The diagnosability of these two contribution methods is further analyzed. Finally, simulation case studies are presented to demonstrate the results.

KW - Concurrent projection to latent structures (CPLS)

KW - Contribution plots

KW - Data-driven

KW - Fault diagnosis

KW - Process monitoring

KW - Quality monitoring

UR - http://www.scopus.com/inward/record.url?scp=84992499976&partnerID=8YFLogxK

U2 - 10.1016/j.ifacol.2015.09.144

DO - 10.1016/j.ifacol.2015.09.144

M3 - Journal Article (refereed)

SN - 2405-8963

VL - 48

SP - 1276

EP - 1281

JO - IFAC-PapersOnLine

JF - IFAC-PapersOnLine

IS - 8

T2 - 9th IFAC Symposium on Advanced Control of Chemical Processes, ADCHEM 2015

Y2 - 7 June 2015 through 10 June 2015

ER -

Fault diagnosis using concurrent projection to latent structures

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this