Structural damage detection under multiple stiffness and mass changes using time series models and adaptive zero-phase component analysis

Nowadays, there is a considerable effort to develop technologies for smart cities. Smart buildings are a critical component of such smart cities, and their automated structural health monitoring is essential. This paper presents a new, efficient, and robust methodology for automated structural damage detection of shear-type buildings. The proposed method uses output-only acceleration response to separately detect changes in stiffness and mass using adaptive zero-phase component analysis (AZCA) in conjunction with time series analysis, that is, autoregressive moving average models with exogenous inputs (ARMAX). In our efforts to tackle the effects of operational factors on structural damage detection processes, herein, mass changes are differentiated from structural damage. Assuming the mass at one DOF at any location is constant (a priori knowledge about the location is not needed), changes in the ARMAX model coefficients are then employed to build stiffness change features (SCFs) and mass change features (MCFs) from which changes in mass and stiffness can be detected separately. A four-story shear structure was constructed in the laboratory to experimentally validate the proposed methodology. The experiment results demonstrate that the approach is successful in eliminating mass effect to determine the existence, location, and severity of the structural damage accurately.