Sideband Algorithm for Automatic Wind Turbine Gearbox Fault Detection and Diagnosis: Preprint

Research output: Contribution to conferencePaper


Improving the availability of wind turbines (WT) is critical to minimize the cost of wind energy, especially for offshore installations. As gearbox downtime has a significant impact on WT availabilities, the development of reliable and cost-effective gearbox condition monitoring systems (CMS) is of great concern to the wind industry. Timely detection and diagnosis of developing gear defectswithin a gearbox is an essential part of minimizing unplanned downtime of wind turbines. Monitoring signals from WT gearboxes are highly non-stationary as turbine load and speed vary continuously with time. Time-consuming and costly manual handling of large amounts of monitoring data represent one of the main limitations of most current CMSs, so automated algorithms are required. This paperpresents a fault detection algorithm for incorporation into a commercial CMS for automatic gear fault detection and diagnosis. The algorithm allowed the assessment of gear fault severity by tracking progressive tooth gear damage during variable speed and load operating conditions of the test rig. Results show that the proposed technique proves efficient and reliable for detecting gear damage.Once implemented into WT CMSs, this algorithm can automate data interpretation reducing the quantity of information that WT operators must handle.
Original languageAmerican English
Number of pages12
StatePublished - 2013
EventEuropean Wind Energy Association 2013 Annual Event - Vienna, Austria
Duration: 4 Feb 20137 Feb 2013


ConferenceEuropean Wind Energy Association 2013 Annual Event
CityVienna, Austria

NREL Publication Number

  • NREL/CP-5000-57395


  • condition monitoring
  • gear tooth fault
  • gearboxes
  • vibration analysis
  • wind turbine


Dive into the research topics of 'Sideband Algorithm for Automatic Wind Turbine Gearbox Fault Detection and Diagnosis: Preprint'. Together they form a unique fingerprint.

Cite this