PCA-based accelerometer data transformation in offshore jacket-type wind turbine support structures for incipient damage detection

Global power capacity is increasingly being comprised by renewable energy sources, where wind farms stand out as paramount power stations. Therefore, the structural health of wind turbines (WTs) represents an essential factor in the energy industry. Specifically, offshore jacket-type WT supports are under critical operational and environmental conditions. Hence, a damage detection strategy is stated, considering several types of structural states and limitations in the quantity of acquired data. The proposed methodology consists of a PCA-based data transformation, in which initially known healthy data are used to be compared with a set of data to be diagnosed; then the damage or healthy states are predicted based on the Mahalanobis distance and threshold value. Because it is a semi-supervised technique, there is no requirement to have damage data on hand to construct the model. The strategy is tested in a scaled-down WT experimental tower.Peer ReviewedObjectius de Desenvolupament Sostenible::7 - Energia Assequible i No ContaminantPostprint (published version