PSI - Issue 78

A. Romanazzi et al. / Procedia Structural Integrity 78 (2026) 777–784

783

(a)

(b)

Fig. 2. Experimental records during seismic input SI18: (a) input-output signals and selection of pre- and post-seismic scenario; and (b) SISO data algorithm segmentation.

(a)

(b)

Fig. 3. Results from the algorithm applied at SI18 records: (a) Bode plot of the frequency response analysis of segmentation; and (b) progression of the first structural frequency along the seismic input. 4. Conclusion A novel approach for dynamic identification of time-variant system based on recursive and iterative ARX data segmentation of non-stationary signals is proposed. The segmentation is performed recursively on the SISO time series by iterating ARX model order in a user-defined range and selecting the best-fitted model according to requirement. Therefore, for each segment, frequency response analysis is conducted considering the selected ARX model as transfer function, and the first frequency is identified in the Bode diagram. The algorithm is validated against single input-single output signals referred to an experimental investigation. The outcomes prove the reliability and robustness of the algorithm to determine the non-linearity of a system along non-stationary signals, and therefore to monitor the evolution of the structural dynamic properties during a seismic event, as the identified frequencies are consistent with the values reported in literature. In conclusion, by comparing the pre- and post-seismic dynamic properties detected by the algorithm, the presented methodology can be applied in emergency scenarios for a fast assessment of structural damage due to seismic events.