In this paper, we introduce a newly developed clustering tool integrated into pyOMA2, an open-source Python module designed for conducting Operational Modal Analysis (OMA). pyOMA2 provides easy access to the most popular algorithms developed over the past two decades, such as Stochastic Subspace Identification (SSI) and Frequency Domain Decomposition (FDD). It supports the analysis of both single and multi-setup data measurements and offers interactive plots and geometry definitions to facilitate the visualization of mode shapes after obtaining modal results. Since version 1.0.1, the software also includes the ability to estimate uncertainty bounds of modal properties for SSI algorithms.

One of the main advantages of pyOMA2 is its modularity, which facilitates the development of additional functionalities. A prime example is the newly developed clustering handler—a specialized class that allows users to define and execute various clustering strategies to automate the selection of modal parameters from SSI results. This tool enables users to implement and compare a large number of the most popular algorithms introduced over the last 15 years, all within the same analysis framework. Furthermore, users have the flexibility to mix specific strategies from different algorithms to tailor a specific clustering process to their needs. Additionally, the integration with the popular machine learning module scikit-learn has expanded the range of available clustering algorithms, providing users with even more options for their analyses. All these capabilities are illustrated in the paper through applications to both a numerical example and real-world datasets.