High frequency vibration spectral monitoring for wind turbine tower structural damage detection

In the framework of NIMO project, a high frequency vibration analysis module has been developed to assess the structural and operational condition of a wind turbine. The high-frequency technique is based on the in-situ measurement and evaluation of vibrations detected in the structure using a dynamic sensor following excitation of the structure by a miniscule mechanical impact. This technique can detect fatigue cracking and structural deterioration due to corrosion by comparing the measurements on a damaged structure with those recorded and calibrated on a defect-free structure. In a practical implementation of high-frequency vibration analysis, the frequencies from 10 kHz up to 200 kHz are analysed. In these frequencies, faults and damages to the structure, even surface damages, are detectable in an early stage. In this paper, results of laboratory tests so-called small-scale test as well as an in-situ full-scale measurements are presented and the efficiency of different numerical identification methods such as (1) spectral cross-correlation, (2) cepstrum analysis, and (3) cumulative normalized power have been examined. Results show that those three spectral analysis methods have different sensitivity for presence of cracks and a combination of the three is much better for crack detection than direct evaluation of the autopower spectrum.