Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • harvard1
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Adaptive Kernel Density-based Anomaly Detection for Nonlinear Systems
Luleå University of Technology, Department of Civil, Environmental and Natural Resources Engineering, Operation, Maintenance and Acoustics.ORCID iD: 0000-0001-7310-5717
Luleå University of Technology, Department of Civil, Environmental and Natural Resources Engineering, Operation, Maintenance and Acoustics.ORCID iD: 0000-0002-7458-6820
Luleå University of Technology, Department of Civil, Environmental and Natural Resources Engineering, Operation, Maintenance and Acoustics.ORCID iD: 0000-0002-0055-2740
Number of Authors: 32018 (English)In: Knowledge-Based Systems, ISSN 0950-7051, E-ISSN 1872-7409, Vol. 139, no 1, p. 50-63Article in journal (Refereed) Published
Abstract [en]

This paper presents an unsupervised, density-based approach to anomaly detection. The purpose is to define a smooth yet effective measure of outlierness that can be used to detect anomalies in nonlinear systems. The approach assigns each sample a local outlier score indicating how much one sample deviates from others in its locality. Specifically, the local outlier score is defined as a relative measure of local density between a sample and a set of its neighboring samples. To achieve smoothness in the measure, we adopt the Gaussian kernel function. Further, to enhance its discriminating power, we use adaptive kernel width: in high-density regions, we apply wide kernel widths to smooth out the discrepancy between normal samples; in low-density regions, we use narrow kernel widths to intensify the abnormality of potentially anomalous samples. The approach is extended to an online mode with the purpose of detecting anomalies in stationary data streams. To validate the proposed approach, we compare it with several alternatives using synthetic datasets; the approach is found superior in terms of smoothness, effectiveness and robustness. A further experiment on a real-world dataset demonstrated the applicability of the proposed approach in fault detection tasks.

Place, publisher, year, edition, pages
Elsevier, 2018. Vol. 139, no 1, p. 50-63
Keywords [en]
maintenance modelling, fault detection, unsupervised learning, nonlinear data, kernel density
National Category
Other Civil Engineering
Research subject
Operation and Maintenance
Identifiers
URN: urn:nbn:se:ltu:diva-60425DOI: 10.1016/j.knosys.2017.10.009ISI: 000417773400005Scopus ID: 2-s2.0-85031504017OAI: oai:DiVA.org:ltu-60425DiVA, id: diva2:1046782
Note

Validerad;2017;Nivå 2;2017-11-21 (andbra)

Available from: 2016-11-15 Created: 2016-11-15 Last updated: 2017-12-28Bibliographically approved
In thesis
1. Big Data Analytics for Fault Detection and its Application in Maintenance
Open this publication in new window or tab >>Big Data Analytics for Fault Detection and its Application in Maintenance
2016 (English)Doctoral thesis, comprehensive summary (Other academic)
Alternative title[sv]
Big Data Analytics för Feldetektering och Applicering inom Underhåll
Abstract [en]

Big Data analytics has attracted intense interest recently for its attempt to extract information, knowledge and wisdom from Big Data. In industry, with the development of sensor technology and Information & Communication Technologies (ICT), reams of high-dimensional, streaming, and nonlinear data are being collected and curated to support decision-making. The detection of faults in these data is an important application in eMaintenance solutions, as it can facilitate maintenance decision-making. Early discovery of system faults may ensure the reliability and safety of industrial systems and reduce the risk of unplanned breakdowns.

Complexities in the data, including high dimensionality, fast-flowing data streams, and high nonlinearity, impose stringent challenges on fault detection applications. From the data modelling perspective, high dimensionality may cause the notorious “curse of dimensionality” and lead to deterioration in the accuracy of fault detection algorithms. Fast-flowing data streams require algorithms to give real-time or near real-time responses upon the arrival of new samples. High nonlinearity requires fault detection approaches to have sufficiently expressive power and to avoid overfitting or underfitting problems.

Most existing fault detection approaches work in relatively low-dimensional spaces. Theoretical studies on high-dimensional fault detection mainly focus on detecting anomalies on subspace projections. However, these models are either arbitrary in selecting subspaces or computationally intensive. To meet the requirements of fast-flowing data streams, several strategies have been proposed to adapt existing models to an online mode to make them applicable in stream data mining. But few studies have simultaneously tackled the challenges associated with high dimensionality and data streams. Existing nonlinear fault detection approaches cannot provide satisfactory performance in terms of smoothness, effectiveness, robustness and interpretability. New approaches are needed to address this issue.

This research develops an Angle-based Subspace Anomaly Detection (ABSAD) approach to fault detection in high-dimensional data. The efficacy of the approach is demonstrated in analytical studies and numerical illustrations. Based on the sliding window strategy, the approach is extended to an online mode to detect faults in high-dimensional data streams. Experiments on synthetic datasets show the online extension can adapt to the time-varying behaviour of the monitored system and, hence, is applicable to dynamic fault detection. To deal with highly nonlinear data, the research proposes an Adaptive Kernel Density-based (Adaptive-KD) anomaly detection approach. Numerical illustrations show the approach’s superiority in terms of smoothness, effectiveness and robustness.

Place, publisher, year, edition, pages
Luleå: Luleå University of Technology, 2016. p. 72
Series
Doctoral thesis / Luleå University of Technology 1 jan 1997 → …, ISSN 1402-1544
Keywords
Big Data analytics, eMaintenance, fault detection, high-dimensional data, stream data mining, nonlinear data
National Category
Other Civil Engineering
Research subject
Operation and Maintenance
Identifiers
urn:nbn:se:ltu:diva-60423 (URN)978-91-7583-769-7 (ISBN)978-91-7583-770-3 (ISBN)
Public defence
2017-01-27, F1031, Luleå University of Technology, 971 87, Luleå, 10:00 (English)
Opponent
Supervisors
Available from: 2016-11-15 Created: 2016-11-15 Last updated: 2018-08-17Bibliographically approved

Open Access in DiVA

fulltext(845 kB)551 downloads
File information
File name FULLTEXT01.pdfFile size 845 kBChecksum SHA-512
6915ba53178929fc852bc1fddb60d9e83dca394cf3618e4c0c3ef8cb922fd6648b9ff2fcea3805894bfb63e3ece620f38450a0030e4516b954d08ceaad1dc6ad
Type fulltextMimetype application/pdf
fulltext(7488 kB)85 downloads
File information
File name FULLTEXT02.pdfFile size 7488 kBChecksum SHA-512
384bd909a0e8d8dcfd56ec22e36c18bb48733ad473502fa74c06161c3d950d053c84709c12ba61efbc7a2cd1935caa829864762c497182260947a14a9a271e64
Type fulltextMimetype application/pdf

Other links

Publisher's full textScopus

Authority records BETA

Zhang, LiangweiLin, JanetKarim, Ramin

Search in DiVA

By author/editor
Zhang, LiangweiLin, JanetKarim, Ramin
By organisation
Operation, Maintenance and Acoustics
In the same journal
Knowledge-Based Systems
Other Civil Engineering

Search outside of DiVA

GoogleGoogle Scholar
Total: 636 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

doi
urn-nbn

Altmetric score

doi
urn-nbn
Total: 814 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • harvard1
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf