Advanced

Performance evaluation of fault detection methods for wastewater treatment processes

Corominas, L.; Villez, K.; Aguado, D.; Rieger, L.; Rosén, Christian LU and Vanrolleghem, P.A. (2011) In Biotechnology and Bioengineering 108(2). p.333-344
Abstract
Several methods to detect faults have been developed in various fields, mainly in chemical and process engineering. However, minimal practical guidelines exist for their selection and application. This work presents an index that allows for evaluating monitoring and diagnosis performance of fault detection methods, which takes into account several characteristics, such as false alarms, false acceptance, and undesirable switching from correct detection to non-detection during a fault event. The usefulness of the index to process engineering is demonstrated first by application to a simple example. Then, it is used to compare five univariate fault detection methods (Shewhart, EWMA, and residuals of EWMA) applied to the simulated results of... (More)
Several methods to detect faults have been developed in various fields, mainly in chemical and process engineering. However, minimal practical guidelines exist for their selection and application. This work presents an index that allows for evaluating monitoring and diagnosis performance of fault detection methods, which takes into account several characteristics, such as false alarms, false acceptance, and undesirable switching from correct detection to non-detection during a fault event. The usefulness of the index to process engineering is demonstrated first by application to a simple example. Then, it is used to compare five univariate fault detection methods (Shewhart, EWMA, and residuals of EWMA) applied to the simulated results of the Benchmark Simulation Model No. 1 long-term (BSM1_LT). The BSM1_LT, provided by the IWA Task Group on Benchmarking of Control Strategies, is a simulation platform that allows for creating sensor and actuator faults and process disturbances in a wastewater treatment plant. The results from the method comparison using BSM1_LT show better performance to detect a sensor measurement shift for adaptive methods (residuals of EWMA) and when monitoring the actuator signals in a control loop (e.g., airflow). Overall, the proposed index is able to screen fault detection methods. Biotechnol. Bioeng. 2011;108: 333–344. © 2010 Wiley Periodicals, Inc. (Less)
Please use this url to cite or link to this publication:
author
organization
publishing date
type
Contribution to journal
publication status
published
subject
keywords
activated sludge, data quality, mathematical modeling, monitoring, process control
in
Biotechnology and Bioengineering
volume
108
issue
2
pages
333 - 344
publisher
John Wiley & Sons
external identifiers
  • scopus:78650219742
ISSN
1097-0290
DOI
10.1002/bit.22953
language
English
LU publication?
yes
id
e8078dbf-a6bd-44bd-8025-dde343c48e7a (old id 4810851)
date added to LUP
2014-11-24 16:06:11
date last changed
2017-01-01 07:31:06
@article{e8078dbf-a6bd-44bd-8025-dde343c48e7a,
  abstract     = {Several methods to detect faults have been developed in various fields, mainly in chemical and process engineering. However, minimal practical guidelines exist for their selection and application. This work presents an index that allows for evaluating monitoring and diagnosis performance of fault detection methods, which takes into account several characteristics, such as false alarms, false acceptance, and undesirable switching from correct detection to non-detection during a fault event. The usefulness of the index to process engineering is demonstrated first by application to a simple example. Then, it is used to compare five univariate fault detection methods (Shewhart, EWMA, and residuals of EWMA) applied to the simulated results of the Benchmark Simulation Model No. 1 long-term (BSM1_LT). The BSM1_LT, provided by the IWA Task Group on Benchmarking of Control Strategies, is a simulation platform that allows for creating sensor and actuator faults and process disturbances in a wastewater treatment plant. The results from the method comparison using BSM1_LT show better performance to detect a sensor measurement shift for adaptive methods (residuals of EWMA) and when monitoring the actuator signals in a control loop (e.g., airflow). Overall, the proposed index is able to screen fault detection methods. Biotechnol. Bioeng. 2011;108: 333–344. © 2010 Wiley Periodicals, Inc.},
  author       = {Corominas, L. and Villez, K. and Aguado, D. and Rieger, L. and Rosén, Christian and Vanrolleghem, P.A.},
  issn         = {1097-0290},
  keyword      = {activated sludge,data quality,mathematical modeling,monitoring,process control},
  language     = {eng},
  number       = {2},
  pages        = {333--344},
  publisher    = {John Wiley & Sons},
  series       = {Biotechnology and Bioengineering},
  title        = {Performance evaluation of fault detection methods for wastewater treatment processes},
  url          = {http://dx.doi.org/10.1002/bit.22953},
  volume       = {108},
  year         = {2011},
}