PSI - Issue 64

6

Mariniello Giulio et al. / Procedia Structural Integrity 64 (2024) 2101–2108 G. Mariniello, D. Coluccino, A. Bilotta, D. Asprone / Structural Integrity Procedia 00 (2019) 000 – 000

2106

To mitigate errors caused by a slight time shift between the signals, the peak of the diagonal measurements is searched for within a time window around the instant when the peak for the vertical measurement is detected. Subsequently, the ratios between the deflection measured by L3 and the variation measured by the inclined sensors are calculated. For each measurement, the correlation coefficients for the k-th measurement are evaluated according to Eq. 3, which defines the ratio between the measured deflection and the variation estimated by j-th sensor. ( , ) = (3) Once these ratios are obtained for each extracted peak, the average correlation coefficient , ̅̅̅̅ is assumed to be the correlation coefficient to test with the other 25% of the non-extracted peaks using Eq.4. , ̂ = , ̅̅̅̅ ∙ (4) where ̂ represents the estimation of the defelction of sensor i through the reading of sensor j To determine how much this estimate deviates from the deflections obtained from the vertical laser, it is possible to calculate a Mean Relative Error (MRE) between measured and estimated displacement. = | ̂( ) − ( )| ( ) (5) where ̂ is the reading obtained from sensor I and i s the instant corresponding to the peaks not selected. 4. Results The approach is evaluated using data recorded by the monitoring system of the Volto Santo Viaduct between April 7 and 9, 2024. These dates encompass a Sunday and a Monday, with markedly different traffic conditions: Sunday is a public holiday with minimal traffic, resulting in few significant displacements, as shown in Fig. 7. In contrast, Monday, a workday, experiences heavier traffic and correspondingly larger displacements.

Fig. 6 Bridge deflection measured by L3 Sensor.