Machine learning methods for wind turbine condition monitoring: A review

Adrian Stetco, Fateme Dinmohammadi, Xingyu Zhao, Valentin Robu, David Flynn, Mike Barnes, John Keane, Goran Nenadic

Research output: Contribution to journalArticle

Abstract

This paper reviews the recent literature on machine learning (ML) models that have been used for condition monitoring in wind turbines (e.g. blade fault detection or generator temperature monitoring). We classify these models by typical ML steps, including data sources, feature selection and extraction, model selection (classification, regression), validation and decision-making. Our findings show that most models use SCADA or simulated data, with almost two-thirds of methods using classification and the rest relying on regression. Neural networks, support vector machines and decision trees are most commonly used. We conclude with a discussion of the main areas for future work in this domain.

Original languageEnglish
Pages (from-to)620-635
Number of pages16
JournalRenewable Energy
Volume133
Early online date9 Oct 2018
DOIs
Publication statusPublished - Apr 2019

Fingerprint

Condition monitoring
Wind turbines
Learning systems
Feature extraction
Decision trees
Fault detection
Turbomachine blades
Support vector machines
Decision making
Neural networks
Monitoring
Temperature

Keywords

  • Condition monitoring
  • Machine learning
  • Prognostic maintenance
  • Renewable energy
  • Wind farms

ASJC Scopus subject areas

  • Renewable Energy, Sustainability and the Environment

Cite this

Stetco, Adrian ; Dinmohammadi, Fateme ; Zhao, Xingyu ; Robu, Valentin ; Flynn, David ; Barnes, Mike ; Keane, John ; Nenadic, Goran. / Machine learning methods for wind turbine condition monitoring: A review. In: Renewable Energy. 2019 ; Vol. 133. pp. 620-635.
@article{28a0f1ec407e43b2bcd302ae58617b10,
title = "Machine learning methods for wind turbine condition monitoring: A review",
abstract = "This paper reviews the recent literature on machine learning (ML) models that have been used for condition monitoring in wind turbines (e.g. blade fault detection or generator temperature monitoring). We classify these models by typical ML steps, including data sources, feature selection and extraction, model selection (classification, regression), validation and decision-making. Our findings show that most models use SCADA or simulated data, with almost two-thirds of methods using classification and the rest relying on regression. Neural networks, support vector machines and decision trees are most commonly used. We conclude with a discussion of the main areas for future work in this domain.",
keywords = "Condition monitoring, Machine learning, Prognostic maintenance, Renewable energy, Wind farms",
author = "Adrian Stetco and Fateme Dinmohammadi and Xingyu Zhao and Valentin Robu and David Flynn and Mike Barnes and John Keane and Goran Nenadic",
year = "2019",
month = "4",
doi = "10.1016/j.renene.2018.10.047",
language = "English",
volume = "133",
pages = "620--635",
journal = "Renewable Energy",
issn = "0960-1481",
publisher = "Elsevier",

}

Machine learning methods for wind turbine condition monitoring: A review. / Stetco, Adrian; Dinmohammadi, Fateme; Zhao, Xingyu; Robu, Valentin; Flynn, David; Barnes, Mike; Keane, John; Nenadic, Goran.

In: Renewable Energy, Vol. 133, 04.2019, p. 620-635.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Machine learning methods for wind turbine condition monitoring: A review

AU - Stetco, Adrian

AU - Dinmohammadi, Fateme

AU - Zhao, Xingyu

AU - Robu, Valentin

AU - Flynn, David

AU - Barnes, Mike

AU - Keane, John

AU - Nenadic, Goran

PY - 2019/4

Y1 - 2019/4

N2 - This paper reviews the recent literature on machine learning (ML) models that have been used for condition monitoring in wind turbines (e.g. blade fault detection or generator temperature monitoring). We classify these models by typical ML steps, including data sources, feature selection and extraction, model selection (classification, regression), validation and decision-making. Our findings show that most models use SCADA or simulated data, with almost two-thirds of methods using classification and the rest relying on regression. Neural networks, support vector machines and decision trees are most commonly used. We conclude with a discussion of the main areas for future work in this domain.

AB - This paper reviews the recent literature on machine learning (ML) models that have been used for condition monitoring in wind turbines (e.g. blade fault detection or generator temperature monitoring). We classify these models by typical ML steps, including data sources, feature selection and extraction, model selection (classification, regression), validation and decision-making. Our findings show that most models use SCADA or simulated data, with almost two-thirds of methods using classification and the rest relying on regression. Neural networks, support vector machines and decision trees are most commonly used. We conclude with a discussion of the main areas for future work in this domain.

KW - Condition monitoring

KW - Machine learning

KW - Prognostic maintenance

KW - Renewable energy

KW - Wind farms

UR - http://www.scopus.com/inward/record.url?scp=85055051302&partnerID=8YFLogxK

U2 - 10.1016/j.renene.2018.10.047

DO - 10.1016/j.renene.2018.10.047

M3 - Article

VL - 133

SP - 620

EP - 635

JO - Renewable Energy

JF - Renewable Energy

SN - 0960-1481

ER -