PSI - Issue 64

Robby Weiser et al. / Procedia Structural Integrity 64 (2024) 492–499 Robby Weiser / Structural Integrity Procedia 00 (2019) 000 – 000

498

7

Figure 6. Example of the one-minute section after classification.

With the help of the TU-BS-1 BiLSTM, entire daily and weekly analyses were carried out in order to be able to make statements about the traffic composition. Figure 7 shows the distribution for 24 h on a working day. It can be seen that the composition of traffic varies over the course of the day. Particularly in the morning and evening hours, the traffic is mainly made up of Group 0 (e.g. cars). The frequency values of trucks in groups 1 and 2 are evenly distributed around the center of the day (12 noon). The analyses are consistent with the observations that rush-hour traffic in particular leads to an increased volume of cars in the morning and evening hours. The consistent alignment between the results and real-world observations serves as both an indication and validation of the classification outcomes which were achieved with the TU-BS-1 BiLSTM.

Figure 7. Distribution for 24 h on a working day.

4. Discussion and outlook In contrast to methods that can only make statements about individual axles by setting up feature vectors Mehdianpour (2002), the TU-BS-1 BiLSTM enables the identification of entire vehicles. This identification of entire vehicle sequences opens up a wide range of applications in measurement data analysis. On the basis of a statistical evaluation, a prediction of future vehicle sequences can be made and a recalculation and service life prediction of a structure can be carried out using realistic load models. In addition, the method presented here can be used to identify the damage-relevant traffic components, in particular from groups 1 and 2 (e.g. trucks). With regard to the retrofitting of the bridges, a repair process caused by welding work under ongoing traffic, heavy goods transport in particular is decisive, as this causes the largest gap openings and has a significant influence on the refurbishment process. The measurement data for the category 1 damage case, which served as the basis for the evaluation in this paper, was located in the immediate vicinity of the deck plate directly under the highly loaded truck lane. Due to the proximity of the crack to the load, global structural influences have less impact on the measurement data than they would have for damage categories in cross girders, cross frames and main girders damage. The stiffness of the components in these damage categories leads to redistribution effects during loading, so that no clear peaks can be identified. As a