On the effectiveness of principal component analysis for decoupling structural damage and environmental effects in bridge structures

An accurate and reliable identification of structural damage is of prime importance to evaluate the structural integrity of civil infrastructure systems. However, the adverse effect of normal fluctuations in the environment on the effectiveness of damage detection techniques remains a continuing challenge in structural health monitoring applications. In this paper, we present the application of principal component analysis (PCA) to temporal damage detection in continuous beam bridge structures subjected to changing environmental effects. For this purpose, we show that sudden discontinuities in the principal components occur at the onset of damage, and that these discontinuities are observed in the projections of the vibration data on the principal components space. The magnitude of the discontinuity is used to define a damage index for damage quantification. A comprehensive numerical study is used to validate the approach on a continuous beam model of highway bridge structures. In particular a sensitivity analysis is conducted to study the effect of both temperature-dependent boundary conditions and material properties on the principal components for multiple damage scenarios. The numerical results show that the approach is robust to mild nonlinearities caused by the effect of temperature on material properties of composite steel-concrete sections and boundary conditions. Furthermore, the approach is experimentally validated using data of the Z24 bridge in Switzerland measured during a period of one year. It is shown that the approach has the capability of tracking the temporal evolution of various damage states induced on the bridge during the testing program.