Data-driven methodologies for change detection in large-scale nonlinear dampers with noisy measurements

Large-scale viscous dampers are frequently used in civil structures to mitigate seismic and wind-induced vibration. For an effective condition assessment of nonlinear dampers, a probabilistic change detection methodology is proposed. The results of experimental studies with different large-scale nonlinear viscous dampers are shown. Considering damper experimental data with measurement uncertainty, the proposed data-driven methodology can be used to (1) detect small changes of a nonlinear system, (2) interpret the physical meanings of system changes, and (3) quantify the uncertainty of the detected changes without a priori knowledge of the system's characteristics.