PSI - Issue 54

7

Francisco Afonso et al. / Procedia Structural Integrity 54 (2024) 545–552 Francisco Afonso / Structural Integrity Procedia 00 (2019) 000 – 000

551

Fig. 5. Railway track defect detection software functioning.

During the software’s analysis, different metrics can be tracked, and visual and audio warnings are enabled, at the same time as a log file is maintained of which and when a specified metric goes outside its limits. Additionally, the metric’s values can be consulted in the difference profile’s graphical area. Five metrics were implemented. Three of those are the same from the previous system: the maximum difference (MD), the average (AVG) and the median (MED). Additionally, horizontal and vertical shifts were included as metrics . The horizontal and vertical shift act on the current acquisition profile, shifting it towards the template profile’s position, vertically and horizontally. The shifts can be applied individually or simultaneously and are calculated using an area of the profiles that is not subject to contact and should not contain any defects from usage. These shifts can be used to detect irregularities in the overall shape and alignment of the wheel. All the metric’s values are in mm. The software incorporates a few more usability-related features, allowing the technician to define a range along the laser line (in mm), where the analysis will take place and to visualize both the sensor ’s FOV and perimeter. The perimeter can be adjusted and is used to filter unwanted interference, only points inside its boundaries will be used in the analysis. 4.3. Trials Repeatability tests were also conducted in a laboratorial setting, a tridimensional model of a wheel containing a gradually increasing surface defect was used. Like the previous system, this model contained seven marks, from -1 to 5, 0 being the last mark before the start of the growing defect, which included the marks 1 to 5. The template profile was acquired in the -1 position and the values of the metrics were registered for each mark; this process was repeated two more times for the same template acquired in the first trial. The results are very similar to the case of the railway tracks monitoring system, where the use of more direct metrics such as the MD results in an error every time the sensor returned to the -1 mark, however, by using the AVG metric, this error was controlled. Finally, some tests were also conducted in a factory environment using an undercarriage which was pushed manually. This system can also be replicated and positioned on the opposing side of the undercarriage, allowing the analysis of both wheels in a single trial.