Structural Nonlinear Damage Detection Based on Time Series Model and Probability Theory

摘要

Damage of practical engineering structure usually possesses nonlinear feature and the results of damage diagnosis are also usually disturbed by different kinds of uncertainty, such as measurement noise and model error. Nonlinear and uncertain characteristics common in measurement signals bring challenges to damage detection. A novel method combined the probability theory with hybrid AR/ARCH model is proposed to detect structural nonlinear damage that couples with measurement uncertainty and structure uncertainty. The AR/ARCH model can fit structural acceleration response time series to effectively extract the nonlinear feature and the calculation probability of damage existence. Probability theory can deal with uncertainty caused by measurement noise or model errors. Therefore, the method based on AR/ARCH model and probability theory can improve the reliability of damage detection compared with deterministic method under different uncertainties. The results of a simulated 5-storey shear structure show the excellent performance to locate nonlinear damage and the potential to quantify the nonlinear damage degree.