Experimental Technology and Management

白噪声激励是振动台试验获取模态参数常用的识别方法，但其长时间激励可能对主频较低（如缩尺超高层）的模型造成轻微损伤，间接影响后续地震波工况抗震性能的计算。针对这一问题，该文在振动台试验的地震波工况中，基于随机子空间法识别了模型的模态参数，并与对应震级后的白噪声工况结果进行对比分析。试验现象及数据分析结果表明，在7度小震、中震及大震的地震波工况下，随机子空间法能够稳定、有效识别缩尺超高层建筑结构模型的模态参数，为同类模型振动台试验的模态参数识别研究提供了方法借鉴。

[Objective] Owing to their extreme height, extended fundamental frequencies, and complex mode shapes, supertall buildings have consistently posed challenges in seismic performance assessments, especially in regions experiencing high seismicity. As a widely adopted experimental approach, shaking table tests provide critical insights into the dynamic response and failure mechanisms of such structures under earthquake loading. A key step in such tests involves the identification of structural modal parameters, which traditionally relies on white noise excitation due to its broadband, uniformly distributed frequency content, and technical maturity. However, for the scaled models of supertall structures, which are typically characterized by reduced stiffness and low natural frequencies, long-duration white noise excitation can induce cumulative microdamage. Such subtle yet potentially irreversible changes, such as local cracking or stiffness degradation, may adversely influence the structural response under subsequent seismic excitations, thereby compromising the accuracy of performance evaluation. [Methods] This study investigates the feasibility of using a stochastic subspace identification(SSI-Dat) method for extracting the modal parameters of a scaled supertall building model directly from seismic response data, thereby avoiding potential model degradation associated with traditional white noise excitation. A series of shaking table tests are conducted under three levels of the seismic input(e.g., minor, moderate, and major earthquakes), corresponding to a seismic fortification intensity of 7. The method utilizes measured acceleration time histories to construct a system observation matrix from which natural frequencies, damping ratios, and dominant mode shapes are extracted via subspace decomposition. To evaluate the reliability and accuracy of the SSI-Dat method, the identified modal parameters are compared with those obtained from white noise excitation using the frequency response function(FRF) method under identical test conditions. [Results] To validate the accuracy and reliability of the SSI-Dat method, the identified modal parameters are compared with those obtained through conventional white noise excitation using the FRF method after each seismic event. Results show that, under minor seismic excitation, the SSI-Dat method yields modal parameters that noticeably deviate from those derived via the white noise approach, likely due to insufficient energy input and limited excitation in high modes. Meanwhile, under moderate and major earthquake inputs, the identified modal parameters, especially fundamental frequencies and dominant mode shapes, exhibit good agreement with the white-noise-based results. This indicates that, as input energy increases, the dynamic response of the structure becomes richer in the modal content, thereby enhancing the effectiveness of subspace-based identification techniques. Throughout the tests, the structure mainly exhibits horizontal motion with minimal torsion. Post-test modal shifts indicate damage accumulation, especially under strong quakes. Despite local stiffness degradation and cracking, the model remains globally intact, retaining load-bearing capacity. This reflects the realistic seismic performance of the supertall building and validates the scaled model's structural fidelity. [Conclusions] This study demonstrates the feasibility and reliability of using the SSI-Dat method to identify the modal parameters of scaled supertall building models under seismic excitation. Compared to the conventional white-noise-based approach, this method avoids risks associated with prolonged artificial excitation and offers a more realistic representation of the structural dynamic characteristics. These findings provide a valuable methodological reference for future shaking table tests involving high-rise structures and contribute to the advancement of nonintrusive modal identification techniques in seismic tests.