Vibration-based system identification of wind turbine system

The goal of this study was to investigate the use of vibration-based system identification methods for detecting wind turbine blade delamination during operating conditions and when subjected to other excitations such as earthquake ground motion. In particular, multivariate singular spectrum analysis and two subspace identification techniques were tested for detecting the dynamic characteristics of the wind blade. A total of two series of experiments was conducted to verify the proposed algorithms. In both cases, accelerometers were installed on the blades for measuring their vibration response. The first test series were performed in the laboratory; a motor spun a smallscale customized wind turbine at controlled angular velocities, and the shaking table excited the entire structure. Artificial damage was introduced by loosening bolts at the blade-rotor connection. The second test was conducted in the field, and vibration data was collected from an operating small-scale wind turbine. The blade's vibration response was analyzed through time-frequency analyses and subspace identification. These tests confirmed that the estimation of dynamic characteristics of blade and rotating frequency of the turbine system was feasible, and the results will guide future monitoring studies planned for larger-scale systems. Copyright (C) 2016 John Wiley & Sons, Ltd.