ARX models for dynamic identification of non-linear systems from non-stationary signals

In the realm of Structural Health Monitoring (SHM), the development of robust and reliable algorithms, combined with new and affordable technologies for sensors and data processing, extended the use of dynamic identification approaches to civil applications. In particular, to support decision making in the reduction of seismic risk, real-time and continuous SHM systems were proposed. However, in common and large cases, e.g. residential buildings, real-time and continuous SHM systems might not be of feasible application due to the large amount of data to be processed and stored. In this framework, a novel algorithm based on the use of Auto-Regressive with eXogenous variable (ARX) models is developed for automated dynamic identification in case of non-linear system and non-stationary signals. Such algorithm is particularly conceived for trigger-based SHM system, in which Single Input-Single Output (SISO) data are recorder only in case of seismic events. Data from experimental shake table tests conducted on a masonry wall are considered to validate the approach, the results of which demonstrate the capability of the algorithm in detecting the frequency shift during a single seismic event.