PSI - Issue 64

Filippo Andreose et al. / Procedia Structural Integrity 64 (2024) 40–47

46

7

Ambient vibrations acquisitions have been processed with a simple filter sequence involving the removal of linear trends, a low-pass filter with a cut-off of 20 Hz, and down-sampling to 40 Hz. FDD and SSI-cov was performed on data acquisitions.

(a)

(b)

(c)

Fig. 8. (a) View of San Faustino bridge, (b) longitudinal geometry, (c) datails of the half-joints.

The first step toward the definition of the subgroup of channels is the preliminary whole bridge modal identification through the FDD algorithm. Since the aim of the analysis was that to evaluate the coupling between the different portions of the bridge, only the principal peaks between frequency of 0 and 8 Hz were picked in the spectrum. Fig. 9a shows the principal results obtained from the FDD analysis. (a) (b)

Fig. 9. Modal results from FDD analysis.

Based on the results of the preliminary FDD analysis, four distinct subgroups were defined, as illustrated in Fig. 9b. The second step is the perform of continuous analysis based on SSI-cov algorithm. Frequency tracking was performed with P3P over period of approximately 4 months, from February 20, 2023, to June 29, 2023. Frequency tracking plots for each subgroup are presented in Fig. 10 (a-d), from which a pretty large success ratio in modal identification points out. Statistical pattern recognition for removal of environmental effects involved proposing a different Multi Linear Regression model for each subgroup. Control charts for all subgroups during the training period are showed in Fig. 10 (e-h). 4. Conclusions The present work illustrated the main features of the P3P software suite for the SHM of existing bridges recently developed within the frame of an ambitious research project involving the largest Italian infrastructure owner and the FABRE consortium. P3P has shown its potential advantage with respect to similar commercial packages in developing real time continuous SHM of a wide set of infrastructure due to its modular framework and high quality post processing tools. The first applications herein described showed promising results that can support further developments in terms of research outcomes as well as in the extension of such approach to the entire target stock of 1000 bridges