PSI - Issue 54

Francisco Afonso et al. / Procedia Structural Integrity 54 (2024) 553–560 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

554

2

1. Introduction As indicated by a survey conducted by Davari et al. (2021), the railway industry can be considered a high operational risk industry, which makes the prevention of failures essential to improve its efficiency and effectiveness. This can be achieved by introducing cyber-physical systems which usually implies the integration of systems and tools via intercommunication systems and intelligent data processing (Davari et al., 2021). These systems can be used in many different contexts within the railway network’s operation . Thus, this work reports on a system developed to monitor railway retaining walls that are located very close to the railway. As explained by Salvini et al. (2013), falling rocks from such walls may result in railway service interruption, damage to the railway infrastructure as well as constituting a dang er to people’s safety. It’s clear that t he failure of retaining walls close to the railway constitutes a serious problem, therefore, it is important to monitor the physical condition of these walls, to enable adequate timely intervention when needed. The system presented in this article was created as part of the overarching project, Ferrovia 4.0, where new tools were developed for implementation in the Portuguese railway network. The focus of these tools was to improve the quality of the railway’s transport service, asset management, safety and passenger comfort. In this case, a system capable of acquiring visual data from retaining walls situated close to the railway tracks was investigated. Several Portuguese railway tracks traverse areas comprised of tall rock and dirt retaining walls. Due to various factors, these walls may decay and, in extreme cases, end up obstructing the railway tracks. The presented system was developed with the sole objective of reconstructing the surface of retaining walls as the train passes by, effectively monitoring their physical condition over time. The retaining wall capture system must adhere to the following requirements: • The system must be implemented in a train • The system’s structure must be easily attachable and removable from its position, without requiring any physical alteration to the train’s structure.

• Obtained data must only pertain to relevant areas of interest • Acquisitions must start and end automatically, using GPS data

Nomenclature fps

Frames per second.

GPS RAM CSV

Global Positioning System Random Access Memory Comma-separated values

2. Solution concept There are several examples of literature written about wall monitoring, especially through the use of photogrammetry. Curtaz et al. (2014) implemented photogrammetric surveys over time to gradually monitor the stability conditions of high-altitude rock blocks in the Mont Blanc Massif, and Salvini et al. (2013) implemented photogrammetric surveys, along with other measurement techniques to reconstruct a tridimensional model of the slope and study its stability. As aforementioned, photogrammetry can be used to measure surfaces, determining their tridimensional positions in space. The surface, on the other hand, must possess sufficient texture in order to allow the detection and matching of points (Luhmann, 2010). There are several methods of implementing photogrammetry in a system to reconstruct a tridimensional surface. One approach is to use a single camera, as described in the work developed by Kaufman et al. (2015), where the