Clustering-based adaptive ground motion selection algorithm for efficient estimation of structural fragilities

This paper proposes a clustering-based algorithm for efficiently selecting a representative subset of ground motions to predict seismic demands of structural systems accurately. The algorithm first identifies critical ground motion features significantly affecting seismic demands using Lasso regression on responses of various single-degree-of-freedom (SDOF) systems. Then, it adaptively selects ground motions based on hierarchical clustering of these features until the fragility curve converges. Applications to reinforced concrete and steel moment-resisting frames demonstrate improved efficiency and wide applicability, confirming robust performance across various ground motions, structural types, and limit-state definitions. The algorithm allows for obtaining structural fragilities using significantly fewer ground motions while maintaining consistency with the initial ground motion set.

Taeyong Kim, Oh-Sung Kwon, Junho Song