Performance of tensor decomposition-based modal identification under nonstationary vibration

摘要: Health monitoring of civil engineering structures is paramount importance when they are subjected to natural hazards or extreme climatic events like earthquake, strong wind gusts man-made excitations. Most the traditional modal identification methods reliant on stationarity assumption vibration response and posed difficulty while analyzing nonstationary (e.g. earthquake human-induced vibration). Recently tensor decomposition based emerged as powerful yet generic blind (i.e. without requiring a knowledge input characteristics) signal tool for structural identification. In this paper, system method further explored estimate parameters using generated due either pedestrian induced excitation in structure. The effects lag sensor densities studied with respect extent nonstationarity responses characterized by stationary duration peak ground acceleration earthquake. A suite more than 1400 earthquakes used investigate performance proposed under wide variety motions utilizing both complete partial measurements high-rise building model. Apart from real-life bridge also verify accuracy method.