Structural Damage Identification Based On GARCH/SV Model

With the development of economy and society, there are more and more large-scale andcomplex structure in our life, the safety of the structure is particularly important. So more andmore experts and scholars pay attention to the related technologies of structural damage detection.With the rapid development of modern technology, technologies of modern sensor, signalacquisition and data processing have been greatly improved. Just under such a background,structural damage identification technology of time series analysis based on structural vibrationinformation is paid more attention by experts and scholars at home and abroad and become a hotarea of research.This thesis summarizes experts and scholars achievement and development trend of the damageidentification. On the basis of understanding time series model theory, Auto Regressive MovingAverage model (ARMA) is applied to extract the linear feature of the acceleration response dataacquired from a three-story building model provided by the LANL USA. Then ConditionalHeteroscedasticity model (GARCH) is applied to conduct time series analysis of the errorsacquired from the three-story building model. The damage sensitive feature is defined according tothe characteristics of the model and damage identification results are good. In order to improve theaccuracy of damage identification results, another nonlinear Stochastic Volatility model (SV) isapplied to conduct time series analysis of the errors acquired from the three-story building model.A new damage sensitive feature is defined and damage identification results are better thanGARCH model’s. The damage identification results get the prospect of research.The innovation of this thesis lays in creatively put forward a new damage sensitive feature ofGARCH model combined with the characteristics of its own, which better to realize nonlineardamage identification. The new damage sensitive feature is the ratio of the standard of theresiduals deviation in the healthy and damage states. The identification results are improved on thebasis of predecessors. The SV model proposed in this thesis is introduced into the area of structuraldamage identification. The thesis proposes SV model procedure to apply for structural damageidentification and gets expected results.