Automatic identification of modal parameters for high arch dams based on SSI incorporating SSA and K-means algorithm

The modal parameters identified by strong seismic observation data of the dam can truly reflect the dynamic characteristics of the structure under working conditions. However, the complex dam- reservoir-foundation structural system under seismic excitation, with dense modalities and strong external disturbances, prevents the accurate identification of the modal parameters of high arch dams on the basis of strong seismic observation data. In this paper, a novel method combining the stochastic subspace identification (SSI), sparrow search algorithm (SSA) and K-means algorithm is proposed to automatically identify the modal parameters of high arch dams. First, based on the strong seismic monitoring data of a high arch dam, SSI is used to identify the modal parameters of the high arch dam, and an improved stabilization diagram is drawn with the natural frequency and modal assurance criterion to avoid the effect of the unstable damping ratio. Second, abnormal poles in the stabilization diagram are identified by combining the local outlier factor and kernel density estimation to eliminate false modes, and the SSA is applied to search for the optimal combination of the K-means clustering algorithm to obtain the initial clustering centers of the stabilization diagram. Finally, the K-means algorithm is used to perform cluster analysis on the effective poles in the stabilization diagram to achieve the automatic identification of the modal parameters. By identifying the modal parameters of a three-degree-of-freedom spring-mass model and a high arch dam under seismic signals, the proposed method is compared with three methods. The results confirm that the novel method is superior to other methods in terms of accuracy and efficiency. The present study can significantly suppress noise, eliminate false modes and automatically identify the real modal parameters without human interference under a low signal-to-noise ratio. & COPY; 2023 Elsevier B.V. All rights reserved.