Identification Of Structral Dynamic Properties And Precdition Of Response Based On Seismic Interferometry

Earthquake is a kind of natural disaster, which is common and destructive. When engineering structures suffer the earthquake, different degrees of damage will appear in structures. Timely detecting the damage can siginficantly reduce loss of life and property. At present, generally adopted method, based on vibration approaches, have difficulties in identifying the damage accurately. This dissertation tries to use seismic interferometry as a tool to identify the related dynamic properties of structures in the linear elastic stage or nonlinear stage. Also, it makes some efforts to predict the response of structures under the effects of earthquakes.Different from the traditional identification methods based on modal analysis, seismic interferometry is a kind of method identifying wave propagating in through the structure. It uses data recorded by accelerometers installed in the structure, computing velocities of the propagating wave. The velocities are mainly dependent on structures themselves. Stiffer structures are, bigger the velocities are. When the structures are damaged, the velocities will change also. So, we can detect the damage by identifying changes of wave velocities. Compared with methods based on modal analysis, seismic interferometry has some favorable features as follows:(1) robustess when applied to system identification;(2) insensitivity to the effects of soil-structures;(3) local in nature. The main contents and conclusions of this dissertation are as follows:(1)After analyzing 3 seismic interferometry methods, the responses of a 20-story frame in 5 earthquakes are studied. Cross correlation and deconvolution(involving direct method and nonlinear least square(LSQ) fit method) are used to identify shear-wave velocites, frequency and damping ratio. In the aspects of identification limitation and stability, deconvolution performs better than cross correlation. In order to obtain responses of nonlinear structures, Open SEES is used to construct the numerical model. Then deconvolution is used to identify changes in wave velocities and frequency. When the peak ground accelration increases from 0.1g to 0.7g, wave velocities attenuate 26% and fundamental frequency attenuates 31%.(2)With the help of impulse response functions, acclerations recorded at the roof are used to compute the response of other floors. IRFs based on deconvolving the recorded response, theoretical IRFs,and LSQ-fitting IRFs are considererd in this dissertation. The first method gets the best results and the error is only 10%...