Abstract: The strain characteristics and crack identification of large-scale prefabricated components during the assembly process in underground metro stations are crucial. Focusing on a prefabricated metro station project, where the distributed fiber optic system (DFOS) is utilized to monitor the strain distribution during the assembly of large-scale components, a principal component analysis-hierarchical clustering (PCA-HC) model is proposed, integrating multiple strain distribution parameters to identify strain distribution features in crack-prone zones. The model′s effectiveness is validated through field tests on slab assembly crack detection. Findings are as follows: (1) The strain features obtained from DFOS indicate average strain growth rates of 105.42% in tension zones and 110.27% in compression zones during roof assembly. The roof assembly process has the most significant effect on the base slab deformation. (2) DFOS clearly reflects the abnormal strain signals at the crack locations. Cracks appear at the contact area between the base slab and the supporting structure during assembly. Increasing the contact area between the support and the assembled components is recommended to mitigate the stress/strain concentration caused by assembly construction. (3) The PCA-HC model effectively identifies potential crack features through dimensionality reduction, feature extraction, and hierarchical clustering, enabling automated crack recognition and classification of large-scale prefabricated components, providing a methodological basis for the safe construction of large-scale prefabricated structures in metro stations.

Key words: metro station, prefabricated assembly structures, distributed fiber optics, crack identification, cluster analysis