Automated physics parameter identification of tuned vibration absorber in offshore wind turbines based on unsupervised spectral clustering and SSI

Abstract

Tuned vibration absorbers (TVBs) are essential for vibration control in offshore wind turbines (OWTs), which are subjected to noticeable vibrations caused by environmental loads. The complex dynamic characteristics of OWT with TVBs pose challenges for traditional methods to accurately identify modal parameters. This paper presents an automated parameter identification method based on spectral clustering and stochastic subspace identification (SSI). This method takes into account the contributions of modal information, with the weighted distance function ensuring the accurate calculation of modal parameter contributions. It effectively improves the accuracy of identifying closely spaced modes. For the numerical simulations, the identification results show high accuracy for the damper when the noise level is below 5.0 %. For the testing model, the average relative deviation in the physics parameters (stiffness and liquid height) of dampers is 2.52 %. This method effectively identifies the physics parameters of TVB, offering theoretical support for the vibration control of OWTs.