Wind turbine state estimation

Research output: Book/ReportReportResearch

657 Downloads (Pure)


Dynamic inflow is an effect which is normally not included in the models used for wind turbine control design. Therefore, potential improvement from including this effect exists. The objective in this project is to improve the methods previously developed for this and especially to verify the results using full-scale wind turbine data. The previously developed methods were based on extended Kalman filtering. This method has several drawback compared to unscented Kalman filtering which has therefore been developed. The unscented Kalman filter was first tested on linear and non-linear test cases which was successful. Then the estimation of a wind turbine state including dynamic inflow was tested on a simulated NREL 5MW turbine was performed. This worked perfectly with wind speeds from low to nominal wind speed as the output prediction errors where white. In high wind where the pitch actuator was always active the results where not as convincing because the output prediction errors where not white. Using real data it has not been possible to get really good results so far. There remains a number of challenges: verifying turbine parameters and getting the most suitable measurement signals, including the 3P effect in the model and perhaps including the 1P effect. It is obviously difficult to make a final conclusion before the above challenges has been resolved.
Original languageEnglish
Number of pages41
Publication statusPublished - 26 Mar 2014


  • Wind energy
  • Dynamic inflow
  • State estimation
  • Kalman filter
  • Unscented transform
  • KF
  • EKF
  • UKF

Fingerprint Dive into the research topics of 'Wind turbine state estimation'. Together they form a unique fingerprint.

Cite this