Output-Only Damage Identification Using Enhanced Structural Characteristic Deflection Shapes and Adaptive Gapped Smoothing Method

Abstract

Structural characteristic deflection shapes (CDSs) such as mode shapes which contain spatial knowledge of structures are highly sensitive for damage detection and localization. Nevertheless, CDSs are vulnerable to measurement noise, which degrades the accuracy of damage identification. In order to enhance CDS-based damage identification, contributions are made in three aspects. First, a robust CDS estimation approach is proposed based on common principal component analysis, which estimates the CDSs as the common diagonalizer of a set of covariance matrices by joint approximation diagonalization (JAD). Second, an adaptive gapped smoothing method (GSM) is proposed and validated to be more accurate than the traditional GSM. Third, a new damage identification index capable of localizing damage and indicating relative damage severity is defined without requiring information of healthy structures. Finally, numerical and experimental examples of beams and a frame with cracks are studied to demonstrate the advantages of the proposed damage identification method in terms of noise robustness and accuracy.