Track condition monitoring in railway applications currently relies on costly and impractical methods, motivating the search for alternatives that use inertial measurements from in-service trains. This thesis investigates one such approach, the Unknown Input Observer, with the objective of assessing whether this observer structure can reconstruct vertical alignment and cross-level geometry in the 3–25~m wavelength range using accelerometer measurements. The Unknown Input Observer was implemented and tested under varying operating conditions on synthetic data generated by a dynamic train model. A second, simpler signal-based method was applied to both real measurement data and synthetically generated data to provide a benchmark for comparison. The results show that the Unknown Input Observer is capable of consistently reconstructing vertical alignment in the targeted wavelength band and cross level for wavelengths above 5~m, under a range of operating conditions at metro-level velocities. The comparison with the signal-based method, however, highlights a potential gap between modeled and real-world performance, underlining the need for further validation with onboard measurement data.
Il monitoraggio delle condizioni della via ferroviaria si basa attualmente su metodi costosi e poco pratici, il che rende necessario sviluppare soluzioni alternative che utilizzino misure inerziali raccolte da treni in servizio. Questa tesi esplora uno di questi approcci, l’Unknown Input Observer, con l’obiettivo di valutare se tale struttura sia in grado di ricostruire l’allineamento verticale e la sopraelevazione nella banda di lunghezze d’onda compresa tra 3 e 25~m a partire da misure accelerometriche. L’Unknown Input Observer è stato implementato e testato in diverse condizioni operative su dati generati da un modello dinamico del treno. Un secondo metodo, più semplice e basato direttamente sulle misure, è stato applicato sia a dati reali sia a dati ottenuti dalle simulazioni per fornire un termine di confronto. I risultati mostrano che l’Unknown Input Observer è in grado di ricostruire in modo affidabile l’allineamento verticale nella banda di lunghezze d’onda considerata e la sopraelevazione per lunghezze superiori a 5~m, in un ampio intervallo di condizioni operative e a velocità tipiche di una metropolitana. Il confronto con il metodo basato sui segnali mette tuttavia in evidenza un potenziale divario tra le prestazioni ottenute in simulazione e quelle reali, sottolineando la necessità di ulteriori validazioni con misure a bordo treno.
The unknown input observer for track geometry reconstruction
MAHIEU, VALENTIJN MARTIN
2025/2026
Abstract
Track condition monitoring in railway applications currently relies on costly and impractical methods, motivating the search for alternatives that use inertial measurements from in-service trains. This thesis investigates one such approach, the Unknown Input Observer, with the objective of assessing whether this observer structure can reconstruct vertical alignment and cross-level geometry in the 3–25~m wavelength range using accelerometer measurements. The Unknown Input Observer was implemented and tested under varying operating conditions on synthetic data generated by a dynamic train model. A second, simpler signal-based method was applied to both real measurement data and synthetically generated data to provide a benchmark for comparison. The results show that the Unknown Input Observer is capable of consistently reconstructing vertical alignment in the targeted wavelength band and cross level for wavelengths above 5~m, under a range of operating conditions at metro-level velocities. The comparison with the signal-based method, however, highlights a potential gap between modeled and real-world performance, underlining the need for further validation with onboard measurement data.| File | Dimensione | Formato | |
|---|---|---|---|
|
2025_10_Mahieu_ExecutiveSummary_02.pdf
accessibile in internet per tutti
Descrizione: Executive summary
Dimensione
1.43 MB
Formato
Adobe PDF
|
1.43 MB | Adobe PDF | Visualizza/Apri |
|
2025_10_Mahieu_Thesis_01.pdf
accessibile in internet per tutti
Descrizione: Thesis
Dimensione
14.05 MB
Formato
Adobe PDF
|
14.05 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/243777