Automatic operational modal analysis for concrete arch dams integrating improved stabilization diagram with hybrid clustering algorithm

Modal identification based on ambient vibration has gained increasing importance in monitoring the operational behavior of dams. This paper develops a robust automated modal identification method to identify the modal parameters of concrete dams. The proposed method requires no modal validation criteria beyond some widely used thresholds. Initially, the covariance-driven stochastic subspace identification (SSI-COV) algorithm is utilized to extract modal parameters, then introducing an improved stabilization diagram to eliminate spurious modes. Subsequently, a hybrid clustering algorithm that combines the clustering by fast search and find of density peaks (DPC) algorithm with a shared nearest neighbor approach is proposed to group physical modes. Clustering centers are determined automatically through a statistics-based method. Finally, the boxplot method is employed to detect and remove outliers from each cluster, thereby facilitating more accurate modal parameter estimation. The performance of the proposed method is validated by identifying the modal parameters of a five-degree-of-freedom frame model and a small-scale arch dam. The results demonstrate that the proposed method is capable of automatically identifying modal parameters with considerable accuracy and robustness.