PSI - Issue 64

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

2105

5

vibration on the supporting system by crossing vehicles. With the aim to evaluate the quality of diagonal layout the correlation between the vertical measure and the diagonal measure is deepened. From a theoretical perspective there is a geometric relationship between the measurement of vertical laser and the measurement of diagonal laser, the ratio between the diagonal sensor and the vertical sensor is equal to the cosine of the angle formed by the vertical with the direction of the diagonal sensor. Nevertheless, only neglecting the major sources of noise due to the support system it is possible to assume that the ratio of vertical to diagonal measurement is constant. Thus, the correlation between vertical and diagonal sensors is analyzed only for the negative peaks of vertical measurement, that are inducted by heavy trucks. Since L3, the vertical laser, represents the target sensor, bridge deflection is determined for each instant concerning the moving average calculated over the 30 s before and following the instant under consideration.

Fig. 5 Measurements collected by L1 sensors compared with heavy load traffic.

In this way, it's possible to mitigate the impact of slow trends related to environmental effects. ( ) = ( ) − 1 ∑ ( )

+30 −30

(1)

Through Eq. 1, it is possible to convert distance measurement to a signal in terms of distance variations, i.e., the bridge deflections. Based on this latter signal, the peaks of the L3 sensor are determined. The peaks identification requires the establishment of a peak threshold (T), all negative displacements above T are collected as a peak. Once, the peaks are identified, 75% of them are randomly selected for estimation of correlation coefficient. The adopted splitting ratio is 0.33 (25:75) that according to Joseph (2022) represent an optimal solution for linear regression. ( ) = ( ) − 1 ∑ ( ) (2)

+30 −30