PSI - Issue 84

Angela Diana et al. / Procedia Structural Integrity 84 (2026) 623–629

627

(a)

(b)

Figure 1. Sensor arrangements: (a) type 1, and (b) type 2.

3.2. Identification In this study, the Covariance-driven Stochastic Subspace Identification (Cov-SSI) algorithm (Tomassini et al., 2025) was employed for the identification of the dynamic characteristics of the analysed infrastructures. This method is widely adopted for OMA due to its robustness in processing noisy signals. Cov-SSI relies on the construction of output covariance matrices and their subsequent decomposition through numerical linear algebra procedures, following the principles highlighted in recent advancements on modal identification applied to densely instrumented structures. The algorithm provides estimates of modal frequencies, mode shapes and damping ratios for different model orders. These estimates are analysed through stabilization criteria to identify the physically meaningful poles. This procedure follows the classical framework of stabilization diagrams commonly used in time-domain system identification, a tool proven to be effective for tracking modal behaviour under variable operational conditions and validated in several benchmark studies, including applications to real bridges (Peeters et al., 2001). The Cov-SSI algorithm was applied to the filtered accelerometric signals described in the previous section, with the objective of identifying the fundamental vibration modes of the monitored spans. In this work, the focus was restricted to the first and second bending frequencies of the decks, for which in previous work predictive equations were established. 4. Results The first and second bending frequencies obtained through the procedure described in the previous sections are reported in Table 1 and their relationship with the span length is shown in Figure 2a-b. Most of the analysed bridges exhibit first bending frequencies (Figure 2a) within a relatively narrow range, between 3.86 Hz and 5.09 Hz. Only a limited number of values fall outside this interval, showing higher frequencies in the range 5.96-7.90 Hz, which are associated with bridges with box-girder decks. Regarding the second bending frequency (Figure 2b), the observed values shift to a higher range, between 13.02 Hz and 18.70 Hz. It is noted that this range does not include all the analysed bridges, as the second bending frequency could not be clearly identified for several structures, including those with box-girder decks.