Wind turbine prognostics

The growing demand for renewable energy has spurred global adoption and expansion of wind turbine technology. To increase energy capacity for existing and future wind farm projects, prognostics and health management (PHM) techniques are considered as a critical approach, whose functionality to diagnose and prognose system degradation and failure is capable to enhance reliability and reduce downtime.

Methods

The Center for Intelligent Maintenance Systems (IMS) established a framework[1] to compute a Global Health Estimator (GHE) to assess real-time turbine performance, and a Local Damage Estimator (LDE), to identify component-level defects.[2]

GHE can be generated from a wind turbine SCADA (Supervisory Control and Data Acquisition) system, by interpreings turbine performance as its capability to generate power under dynamic environmental conditions. Wind speed, wind direction, pitch angle and othera parameters are first selected as input. Then two key parameters in characterizing wind power generation, wind speed and actual power output, collected while turbine is known to work under nominal healthy condition are used to establish a baseline model. When real-time data arrives, same parameters are used to model current performance. GHE is obtained by computing the distance between the new data and its baseline model.

By trending the GHE over time, performance prediction can be made when unit revenue will drop below a predetermined break-even threshold. Maintenance should be triggered and directed to components with low LDE values. LDE is computed based on measurements from condition monitoring system (CMS) and SCADA, and is used to locate and diagnose incipient failure at component level.

Machine learning is also used by collecting and analyzing massive amounts of data such as vibration, temperature, power and others from thousands of wind turbines several times per second to predict and prevent failures.[3]

Case studies

Performance assessment based on GHE is validated with a case study from an onshore wind turbine with 26 months of data. The results show a degrading trend of GHE that lead to major turbine downtime periods before which predictive maintenance could be made.[4]

Drive train prognostics is validated with SCADA and CMS data from an offshore turbine.[5]

Demonstrations

A Wind Turbine Health Monitoring System based on SCADA and CMS data is shown at the WindturbinePHM.com[6] website.[7] Additional discussion on the uses of wind farm SCADA data for predictive maintenance and performance improvement can be found on the SCADA Miner website.[8]

See also

References

  1. ↑ "Center for Intelligent Maintenance Systems — IMS Center". Imscenter.net. Retrieved 2014-06-04.
  2. ↑ "Wind Turbine Prognostics and Health Management — IMS Center". Windturbinephm.com. 2012-05-01. Retrieved 2014-06-04.
  3. ↑ "Neurale netværk kan forudsige, hvornår møllens tandhjul knækker". Version2/Ingeniøren. Retrieved 19 November 2016.
  4. ↑ "Wind turbine performance assessment using multi-regime modeling approach". Renewable Energy. 45: 86–95. 2012-09-30. doi:10.1016/j.renene.2012.02.018. Retrieved 2014-06-04.
  5. ↑ Wenyu Zhao; David Siegel; Jay Lee; Liying Su. "An Integrated Framework of Drivetrain Degradation Assessment and Fault Localization for Offshore Wind Turbines" (PDF). Retrieved August 20, 2016.
  6. ↑ "Wind Turbine Prognostics and Health Management — IMS Center".
  7. ↑ "IMS Wind Turbine PHM Demonstration — IMS Center". Imscenter.net. 2012-06-14. Retrieved 2014-06-04.
  8. ↑ "Wind Farm SCADA Data Analysis - SCADA Miner". Retrieved 2016-09-26.
This article is issued from Wikipedia. The text is licensed under Creative Commons - Attribution - Sharealike. Additional terms may apply for the media files.