Offshore wind power stands as a clean and low-carbon energy option that is booming as part of the efforts to achieve the goal of carbon neutrality. Effectively monitoring the dynamic response of wind turbines is a necessity to analyze the modal parameters, which are key parameters to assess whether the wind turbines are operating safely. Modal parameter identification for offshore wind turbines (OWTs) becomes essential through analyzing the dynamic response, given the limited acceptable range of natural frequencies under dynamic loads. This paper introduces a novel machine learning-based method that combines the SSI-data (data-driven stochastic subspace identification) modal parameter identification method with clustering analysis, employing DBSCAN (Density-Based Spatial Clustering of Applications with Noise) and the K-means cluster algorithm. The proposed method can automatically define the number of K-means clusters. The validation was carried out through a theoretical analysis using a four-degree-of-freedom model and Opensees numerical simulation model of an OWT. The verification and case study outcomes demonstrate that the proposed method possesses the accuracy required for automated modal parameter identification. Compared with the benchmark case results, the differences between the frequencies identified by the proposed method and the reference values are 0.0%, 0.30%, and 0.18% for the first three orders, respectively. This research not only provides valuable insights for professionals in related dynamic monitoring fields but also offers technical support for diagnosing abnormal states of OWTs utilizing dynamic response data.
Yang et al. (2026) studied this question.