Analysis of Acoustic Emission Data from Wind Turbine Blade Testing Using Unsupervised Pattern Recognition


15th International Acoustic Emission Symposium, Tokyo, Japan, 11-14 Sept. 2000
A. A., Anastassopoulos, S. J., Vahaviolos, D. A. Kouroussis, P. Vionis, J. C. Lenain, A. Proust


Acoustic Emission testing of FRP structures, such as Wind Turbine blades, is a challenging application because the blades themselves are of complex design, and FRP materials are, by nature, emissive, when subjected to loading. During loading of a blade, different damage mechanisms, which produce AE, might be initiated or ceased at different load levels and at different sections of the blade, or even coexist. Current Wind Turbine blade certification practices are based solely on visual inspection and heuristic quantification of audible damage indications. Therefore, segregation of the damage mechanisms and characterization of their criticality with load, by means of AE, is essential for both the structural integrity assessment of the blade and the understanding of the damage evolution with increasing load. In the present work, similar Wind Turbine blades were tested with AE, at different loading envelopes and load levels. Unsupervised Pattern Recognition analysis was performed on the corresponding Acoustic Emission data which were clustered by UPR, based on their AE characteristics. The resulting clusters of data were compared with respect to their criticality, AE features and location on the blade. Particular clustering algorithms were proved to be very efficient in discriminating the various AE mechanisms for each test case, while clusters with similar AE characteristics appeared in different tests. Overall, UPR analysis proved to be a powerful tool towards the evaluation and physical interpretation of AE data.