Adaptive training of vibration-based anomaly detector for wind turbine condition monitoring

Takanori Hasegawa, Jun Ogata, Masahiro Murakawa, Tetsunori Kobayashi, Tetsuji Ogawa

研究成果: Conference contribution

1 引用 (Scopus)

抄録

Adaptive training of a vibration-based anomaly detector forwind turbine condition monitoring system (CMS) is carriedout to achieve high-performance detection from the earlystages of monitoring. Machine learning-based wind turbineCMSs are required to collect large-scale data to yield reliablepredictions. Existing studies in this area have postulatedthat both data for training a monitoring system and those duringthe operation of the system are obtained from identicaldevices. In addition, constant monitoring of data is desirable,but in practice, the data can be observed periodically(e.g., several tens of seconds of data are observed every twohours). In this case, collecting sufficient data is time consuming,making it difficult to conduct accurate predictions atthe early stage of the CMS operation. To address this problem,a small amount of vibration data observed at a targetwind turbine is utilized to adapt the anomaly detector thatis trained on relatively large-scale vibration signals obtainedfrom other wind turbines. In the present study, maximum aposteriori (MAP) adaptation is applied to a Gaussian mixturemodel (GMM)-based anomaly detector. Experimentalcomparisons using vibration data from the gearbox in the experimentalenvironment and those used in the wind turbinedemonstrated that MAP-based GMM adaptation yielded animprovement in anomaly detection accuracy even when onlya small amount of data is observed at the target gearbox.

元の言語English
ホスト出版物のタイトルPHM 2017 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2017
編集者Matthew J. Daigle, Anibal Bregon
出版者Prognostics and Health Management Society
ページ177-184
ページ数8
ISBN(電子版)9781936263059
出版物ステータスPublished - 2017 1 1
イベント9th Annual Conference of the Prognostics and Health Management Society, PHM 2017 - St. Petersburg, United States
継続期間: 2017 10 22017 10 5

出版物シリーズ

名前Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM
ISSN(印刷物)2325-0178

Conference

Conference9th Annual Conference of the Prognostics and Health Management Society, PHM 2017
United States
St. Petersburg
期間17/10/217/10/5

Fingerprint

Condition monitoring
Vibration
Wind turbines
Detectors
Monitoring
Turbines
Learning systems

ASJC Scopus subject areas

  • Information Systems
  • Electrical and Electronic Engineering
  • Health Information Management
  • Computer Science Applications

これを引用

Hasegawa, T., Ogata, J., Murakawa, M., Kobayashi, T., & Ogawa, T. (2017). Adaptive training of vibration-based anomaly detector for wind turbine condition monitoring. : M. J. Daigle, & A. Bregon (版), PHM 2017 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2017 (pp. 177-184). (Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM). Prognostics and Health Management Society.

Adaptive training of vibration-based anomaly detector for wind turbine condition monitoring. / Hasegawa, Takanori; Ogata, Jun; Murakawa, Masahiro; Kobayashi, Tetsunori; Ogawa, Tetsuji.

PHM 2017 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2017. 版 / Matthew J. Daigle; Anibal Bregon. Prognostics and Health Management Society, 2017. p. 177-184 (Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM).

研究成果: Conference contribution

Hasegawa, T, Ogata, J, Murakawa, M, Kobayashi, T & Ogawa, T 2017, Adaptive training of vibration-based anomaly detector for wind turbine condition monitoring. : MJ Daigle & A Bregon (版), PHM 2017 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2017. Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM, Prognostics and Health Management Society, pp. 177-184, 9th Annual Conference of the Prognostics and Health Management Society, PHM 2017, St. Petersburg, United States, 17/10/2.
Hasegawa T, Ogata J, Murakawa M, Kobayashi T, Ogawa T. Adaptive training of vibration-based anomaly detector for wind turbine condition monitoring. : Daigle MJ, Bregon A, 編集者, PHM 2017 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2017. Prognostics and Health Management Society. 2017. p. 177-184. (Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM).
Hasegawa, Takanori ; Ogata, Jun ; Murakawa, Masahiro ; Kobayashi, Tetsunori ; Ogawa, Tetsuji. / Adaptive training of vibration-based anomaly detector for wind turbine condition monitoring. PHM 2017 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2017. 編集者 / Matthew J. Daigle ; Anibal Bregon. Prognostics and Health Management Society, 2017. pp. 177-184 (Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM).
@inproceedings{88757ba5df3e4e1f8cbd54d683cc87de,
title = "Adaptive training of vibration-based anomaly detector for wind turbine condition monitoring",
abstract = "Adaptive training of a vibration-based anomaly detector forwind turbine condition monitoring system (CMS) is carriedout to achieve high-performance detection from the earlystages of monitoring. Machine learning-based wind turbineCMSs are required to collect large-scale data to yield reliablepredictions. Existing studies in this area have postulatedthat both data for training a monitoring system and those duringthe operation of the system are obtained from identicaldevices. In addition, constant monitoring of data is desirable,but in practice, the data can be observed periodically(e.g., several tens of seconds of data are observed every twohours). In this case, collecting sufficient data is time consuming,making it difficult to conduct accurate predictions atthe early stage of the CMS operation. To address this problem,a small amount of vibration data observed at a targetwind turbine is utilized to adapt the anomaly detector thatis trained on relatively large-scale vibration signals obtainedfrom other wind turbines. In the present study, maximum aposteriori (MAP) adaptation is applied to a Gaussian mixturemodel (GMM)-based anomaly detector. Experimentalcomparisons using vibration data from the gearbox in the experimentalenvironment and those used in the wind turbinedemonstrated that MAP-based GMM adaptation yielded animprovement in anomaly detection accuracy even when onlya small amount of data is observed at the target gearbox.",
author = "Takanori Hasegawa and Jun Ogata and Masahiro Murakawa and Tetsunori Kobayashi and Tetsuji Ogawa",
year = "2017",
month = "1",
day = "1",
language = "English",
series = "Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM",
publisher = "Prognostics and Health Management Society",
pages = "177--184",
editor = "Daigle, {Matthew J.} and Anibal Bregon",
booktitle = "PHM 2017 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2017",

}

TY - GEN

T1 - Adaptive training of vibration-based anomaly detector for wind turbine condition monitoring

AU - Hasegawa, Takanori

AU - Ogata, Jun

AU - Murakawa, Masahiro

AU - Kobayashi, Tetsunori

AU - Ogawa, Tetsuji

PY - 2017/1/1

Y1 - 2017/1/1

N2 - Adaptive training of a vibration-based anomaly detector forwind turbine condition monitoring system (CMS) is carriedout to achieve high-performance detection from the earlystages of monitoring. Machine learning-based wind turbineCMSs are required to collect large-scale data to yield reliablepredictions. Existing studies in this area have postulatedthat both data for training a monitoring system and those duringthe operation of the system are obtained from identicaldevices. In addition, constant monitoring of data is desirable,but in practice, the data can be observed periodically(e.g., several tens of seconds of data are observed every twohours). In this case, collecting sufficient data is time consuming,making it difficult to conduct accurate predictions atthe early stage of the CMS operation. To address this problem,a small amount of vibration data observed at a targetwind turbine is utilized to adapt the anomaly detector thatis trained on relatively large-scale vibration signals obtainedfrom other wind turbines. In the present study, maximum aposteriori (MAP) adaptation is applied to a Gaussian mixturemodel (GMM)-based anomaly detector. Experimentalcomparisons using vibration data from the gearbox in the experimentalenvironment and those used in the wind turbinedemonstrated that MAP-based GMM adaptation yielded animprovement in anomaly detection accuracy even when onlya small amount of data is observed at the target gearbox.

AB - Adaptive training of a vibration-based anomaly detector forwind turbine condition monitoring system (CMS) is carriedout to achieve high-performance detection from the earlystages of monitoring. Machine learning-based wind turbineCMSs are required to collect large-scale data to yield reliablepredictions. Existing studies in this area have postulatedthat both data for training a monitoring system and those duringthe operation of the system are obtained from identicaldevices. In addition, constant monitoring of data is desirable,but in practice, the data can be observed periodically(e.g., several tens of seconds of data are observed every twohours). In this case, collecting sufficient data is time consuming,making it difficult to conduct accurate predictions atthe early stage of the CMS operation. To address this problem,a small amount of vibration data observed at a targetwind turbine is utilized to adapt the anomaly detector thatis trained on relatively large-scale vibration signals obtainedfrom other wind turbines. In the present study, maximum aposteriori (MAP) adaptation is applied to a Gaussian mixturemodel (GMM)-based anomaly detector. Experimentalcomparisons using vibration data from the gearbox in the experimentalenvironment and those used in the wind turbinedemonstrated that MAP-based GMM adaptation yielded animprovement in anomaly detection accuracy even when onlya small amount of data is observed at the target gearbox.

UR - http://www.scopus.com/inward/record.url?scp=85062842600&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85062842600&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:85062842600

T3 - Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM

SP - 177

EP - 184

BT - PHM 2017 - Proceedings of the Annual Conference of the Prognostics and Health Management Society 2017

A2 - Daigle, Matthew J.

A2 - Bregon, Anibal

PB - Prognostics and Health Management Society

ER -