A new method for wind turbine drive-train condition monitoring is proposed: the innovative idea is that vibrations are measured at the tower. The critical point is extracting knowledge about the drive-train from tower measurements: this is achieved by measuring simultaneously at the highest possible number of nearby wind turbines. One wind turbine is selected as target and the others are used as reference. The data are analyzed in the time domain basing on statistical features (root mean square, peak, crest factor, skewness, kurtosis). The data set in the feature space reduces to a matrix, from which the observations at the target wind turbine should be distinguishable. The application of this algorithm is supported by univariate statistical tests and by Principal Component Analysis. A novelty index based on the Mahalanobis distance is finally used to detect the statistical novelty of the damaged wind turbine. This work is based on field measurement campaigns, performed by the authors in 2018 and 2019 at wind farms owned by the Renvico company.
Wind turbine drive-train condition monitoring through tower vibrations measurement and processing
Astolfi D.;
2020-01-01
Abstract
A new method for wind turbine drive-train condition monitoring is proposed: the innovative idea is that vibrations are measured at the tower. The critical point is extracting knowledge about the drive-train from tower measurements: this is achieved by measuring simultaneously at the highest possible number of nearby wind turbines. One wind turbine is selected as target and the others are used as reference. The data are analyzed in the time domain basing on statistical features (root mean square, peak, crest factor, skewness, kurtosis). The data set in the feature space reduces to a matrix, from which the observations at the target wind turbine should be distinguishable. The application of this algorithm is supported by univariate statistical tests and by Principal Component Analysis. A novelty index based on the Mahalanobis distance is finally used to detect the statistical novelty of the damaged wind turbine. This work is based on field measurement campaigns, performed by the authors in 2018 and 2019 at wind farms owned by the Renvico company.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
