Structural Damage Identification Based On Conditional Heteroscedasticity Time Series Model

Structural damage identification is one of very active research subjects in current civil engineering field, having extensive and practical engineering application background, and the related theories and technologies also are developing constantly. As sensor technologies, signal acquiring technologies and system recognition technologies are developing rapidly; structural damage identification based on dynamic responses has become a hot topic in health monitoring and damage detection research. Structural damage identification based on time series methods has got more and more considerable attention from experts at home and abroad. The term time series has originated in statistics then becomes an indispensable and important component in a system identification field,It has successfully applied in structural damage identification which it is a new structural damage identification method and would leading the frontier in the field.This paper reviewed several categories existing structural damage identification methods based on time series analysis, and puts forward a classical linear time series model ARMA which successfully realize structural nonlinear damage identification. A damage sensitive feature has been proposed in this linear time series model. Subsequently, nonlinear time series model GARCH has been presented to solve some limitations which display in structural nonlinear damage identification using linear time series model ARMA. The feasibility, validity and applicability of the two methods have been tested on the analytical and experimental results of a three-story building structure which performs nonlinear damage situation. By comparing the linear model ARMA and nonlinear model GARCH, some conclusions have been made that the nonlinear time series model GARCH showed its superiority of nonlinear damage identification in small damage and resistance to environmental factor influence and so on. The innovation of this paper lies in establishing the simple, classic linear model ARMA to achieve structural nonlinear damage identification, and proposes the nonlinear time series model GARCH for the first time, which better to realize nonlinear damage identification. The nonlinear conditional heteroscedastic GARCH model proposed in this paper creates a novel nonlinear time series analysis methodology in the application of structural damage identification.