Offshore substation serves as a pivotal structure in offshore wind farms. However, its harsh operational environment exposes it to unpredictable safety issues under seismic motions. In order to clarify the damage status of the upper structure of the offshore substation in the earthquake, it is necessary to systematically identify the structure. This study establishes a finite element (FE) model of an offshore substation, uses typical seismic waves as input to obtain the acceleration time history responses at the beam-column joints, and adopts the data-driven stochastic subspace identification (SSI) method to identify the structural modal parameters. The identified natural frequencies and modes are compared with the modal parameters obtained by finite element analysis (FEA). The results show that the data-driven SSI method can accurately estimate the modal parameters of the structure under unknown input conditions. Under seismic loads, the SSI method can identify global modes without requiring a high model order, providing an effective method for the vibration monitoring of offshore wind farms.