Detection of seismic damage in buildings using time series analysis and pattern recognition

In seismically active regions, the ability to accurately and efficiently quantify seismic-induced damage to civil infrastructure is imperative. This study focuses on the use of time series analysis techniques in SHM. Autoregressive (AR) models were applied to the acceleration time histories, obtained by computer simulations, of a simple 3-storey building subjected to random ground motion. Damage in the structure was simulated by a reduction in lateral stiffness. The proposed method was shown to provide good damage estimations and to be unaffected by small amounts of noise and changes in operating conditions.