Damage identification of wind turbine's blades using power spectral density and kriging analysis

Small faults, cracks, and other structural defects that occur in some components of a wind turbine, might lead to a catastrophic failure. Moreover, an important operating requirement that relates to a wind turbines airfoils are its ability to perform when the smoothness of its surface is compromised. The accreted dust on the surface of blade increases the drag of the airfoil and a decrease in the lift, while large accumulation can lead to complete turbine stops, thereby reduction in the power output of the wind turbine. Additional, to accreted dust and debris. To prevent such failures, proactive measures have to be taken to identify and detect defects at its early stages. In this paper, vibration signature of the structure is utilized for identification and detection of defects. Changes in resonant frequencies and resonant amplitude of the turbine blades are compared before and after damage. These changes are utilized as means for identifying damage in the blades. A 2k factorial experiment is constructed to generate changes in resonant frequencies and spectral amplitudes due to changes in crack length, location from the center of the blades' hub, and the orientation of the crack. Three accelerometers placed at the hub center, middle and tip of the blade measure the acceleration at corresponding locations. Power spectral density (PSD) of acceleration is generated for various test conditions in the factorial experiment. Damage in the vicinity of the accelerometers locations have well defined power spectral densities. However damage characteristics at all other locations are predicted using the Kriging method in which, given measurements at a set of locations in a region, Kriging creates a map of predicted value throughout the region. Damage characteristics estimates using the proposed method revealed an error as low as 0.3%. Simulation is used to validate the proposed method and the results are discussed.