Singular spectrum analysis based structural damage detection from nonlinear vibration measurements containing noise

Abstract

Singular spectrum analysis is developed for the processing of nonlinear vibration measurements containing noise in structural damage detection. The method uses system augmentation to make the nonlinear system into an augmented linear system. The measured response of the system is decomposed and the corresponding decomposition subspace of the components which contains the least interference is reconstructed. Direct system parameter identification technique is then used to solve the model properties of the augmented system. Minimum rank perturbation theory is generalized so that it can be applied to detect local damage of the augmented system. Excellent identification results are obtained when Singular Spectrum Analysis (SSA) is applied, while identification result that without applying SSA shows large errors. The enhanced method is shown to be capable of yielding accurate identified results with noisy measurement in nonlinear systems.