Prediction of wind turbine generator bearing failure through analysis of high frequency vibration data and the application of support vector machine algorithms

Research output: Contribution to conferencePaper

Abstract

Reducing wind turbine downtime through innovations surrounding asset management has the potential to greatly influence the overall levelised cost of energy of large onshore and offshore developments. This research paper uses multiple examples of the same generator bearing failure to provide insight into how condition monitoring systems can be used in to train machine learning algorithms with the ultimate goal of predicting failure and remaining useful life. Results show that by analysing high frequency vibration data and extracting key features to train support vector machine algorithms, an accuracy of 67% can be achieved in successfully predicting failure 1-2 months before occurrence. This paper reflects on the limitations surrounding a generalised training approach, taking advantage of all available data, showing that if too many different examples are considered of different wind turbines and operating conditions the overall accuracy can be diminished.

Conference

ConferenceThe 7th International Conference on Renewable Power Generation
Abbreviated titleRPG2018
CountryDenmark
CityCopenhagen
Period26/08/1827/09/18
Internet address

Fingerprint

Bearings (structural)
Turbogenerators
Wind turbines
Support vector machines
Asset management
Condition monitoring
Learning algorithms
Learning systems
Innovation
Costs

Keywords

  • wind turbine
  • generator
  • failure
  • bearing
  • condition monitoring

Cite this

@conference{7273fc89c3f74a079ca64c12aba54221,
title = "Prediction of wind turbine generator bearing failure through analysis of high frequency vibration data and the application of support vector machine algorithms",
abstract = "Reducing wind turbine downtime through innovations surrounding asset management has the potential to greatly influence the overall levelised cost of energy of large onshore and offshore developments. This research paper uses multiple examples of the same generator bearing failure to provide insight into how condition monitoring systems can be used in to train machine learning algorithms with the ultimate goal of predicting failure and remaining useful life. Results show that by analysing high frequency vibration data and extracting key features to train support vector machine algorithms, an accuracy of 67{\%} can be achieved in successfully predicting failure 1-2 months before occurrence. This paper reflects on the limitations surrounding a generalised training approach, taking advantage of all available data, showing that if too many different examples are considered of different wind turbines and operating conditions the overall accuracy can be diminished.",
keywords = "wind turbine, generator, failure, bearing, condition monitoring",
author = "A. Turnbull and J. Carroll and S. Koukoura and A. McDonald",
year = "2018",
month = "9",
day = "28",
language = "English",
note = "The 7th International Conference on Renewable Power Generation , RPG2018 ; Conference date: 26-08-2018 Through 27-09-2018",
url = "https://events.theiet.org/rpg/index.cfm",

}

Prediction of wind turbine generator bearing failure through analysis of high frequency vibration data and the application of support vector machine algorithms. / Turnbull, A.; Carroll, J.; Koukoura, S.; McDonald, A.

2018. Paper presented at The 7th International Conference on Renewable Power Generation , Copenhagen, Denmark.

Research output: Contribution to conferencePaper

TY - CONF

T1 - Prediction of wind turbine generator bearing failure through analysis of high frequency vibration data and the application of support vector machine algorithms

AU - Turnbull, A.

AU - Carroll, J.

AU - Koukoura, S.

AU - McDonald, A.

PY - 2018/9/28

Y1 - 2018/9/28

N2 - Reducing wind turbine downtime through innovations surrounding asset management has the potential to greatly influence the overall levelised cost of energy of large onshore and offshore developments. This research paper uses multiple examples of the same generator bearing failure to provide insight into how condition monitoring systems can be used in to train machine learning algorithms with the ultimate goal of predicting failure and remaining useful life. Results show that by analysing high frequency vibration data and extracting key features to train support vector machine algorithms, an accuracy of 67% can be achieved in successfully predicting failure 1-2 months before occurrence. This paper reflects on the limitations surrounding a generalised training approach, taking advantage of all available data, showing that if too many different examples are considered of different wind turbines and operating conditions the overall accuracy can be diminished.

AB - Reducing wind turbine downtime through innovations surrounding asset management has the potential to greatly influence the overall levelised cost of energy of large onshore and offshore developments. This research paper uses multiple examples of the same generator bearing failure to provide insight into how condition monitoring systems can be used in to train machine learning algorithms with the ultimate goal of predicting failure and remaining useful life. Results show that by analysing high frequency vibration data and extracting key features to train support vector machine algorithms, an accuracy of 67% can be achieved in successfully predicting failure 1-2 months before occurrence. This paper reflects on the limitations surrounding a generalised training approach, taking advantage of all available data, showing that if too many different examples are considered of different wind turbines and operating conditions the overall accuracy can be diminished.

KW - wind turbine

KW - generator

KW - failure

KW - bearing

KW - condition monitoring

M3 - Paper

ER -