PSI - Issue 84

Antonio S. López-Cuervo et al. / Procedia Structural Integrity 84 (2026) 223–230

225

Fig. 1(b). A multi-setup strategy was adopted (two sensor setups per AVT) to provide dense coverage of the main truss. For each setup, 13-min records were acquired at a sampling frequency of 1600 Hz. The signals were pre processed by detrending and spike removal, followed by a 4th-order Butterworth band-pass filter (2.2–100 Hz) and decimation to 200 Hz for subsequent analysis. Figures 2(a) and 2(b) show the filtered time series from a representative channel for AVT-Acc-1 and AVT-Acc-2, respectively.

Fig. 1. Steel footbridge and monitoring system: (a) general view of the main Warren truss, (b) accelerometer installed, (c) FE model, (d) Wheatstone bridge and ADC for strain measurement, and (e) strain gauge installed.

Fig. 2. Filtered time series measured at a representative channel for (a) AVT-Acc-1, (b) AVT-Acc-2, (c) AVT-SG-1, and (d) AVT-SG-2.

For the SG-based AVTs, FLAB-6-11 resistive strain gauges from TML were installed (Fig. 1(e)) and wired in a quarter-bridge configuration. The output of each Wheatstone bridge was connected to an HX711 module (24-bit ADC, gain 128), and the circuitry was housed in a sealed enclosure for field deployment (Fig. 1(d)). The circuit was connected via shielded Ethernet cables to an Arduino Mega microcontroller, used for data acquisition. Record lengths