This thesis contributes to the development of a diagnostic algorithm for detecting wheel defects—specifically, wheel-flat defects—in rail systems, paving the way for integration into predictive maintenance protocols. Wheel-flat defects are discrete irregularities in wheel geometry, typically resulting from sliding during braking or insufficient rail adhesion, which produce periodic vibrations and accelerate wear on both rolling stock and infrastructure. An advanced sensor node was designed to acquire axle-box vertical acceleration data, from which key features were extracted. The system was validated through a test campaign on the San Donato circuit in Bologna, Italy. Sensitivity analyses examined how defect dimensions, the number of defects, vehicle speed, track layout, and axle load affect diagnostic performance. A binary classification algorithm was then developed to distinguish between defective and non-defective wheel states. The results confirm the effectiveness of the proposed approach and its potential to enhance condition-based maintenance in the rail industry.
Questa tesi contribuisce allo sviluppo di un algoritmo per il rilevamento dei difetti delle ruote—in particolare, dei difetti di tipo wheel-flat—nei sistemi ferroviari, aprendo la strada all’integrazione in protocolli di manutenzione predittiva. I difetti wheel-flat rappresentano irregolarità discrete nella geometria della ruota, tipicamente causate dallo slittamento durante la frenata o da un’insufficiente aderenza con il binario, che generano vibrazioni periodiche e accelerano l’usura sia del materiale rotabile che dell’infrastruttura. È stato progettato un sensore per acquisire i dati di accelerazione verticale in boccola, da cui sono state estratte le principali feature. Il sistema è stato validato attraverso una campagna sperimentale sul circuito di San Donato a Bologna, in Italia. Analisi di sensibilità hanno esaminato l’influenza delle dimensioni del difetto, del numero di difetti, della velocità del veicolo, del layout del tracciato e del carico assiale sulle prestazioni diagnostiche. Successivamente, è stato sviluppato un algoritmo di classificazione binaria in grado di distinguere tra ruote con difetto e ruote senza difetto. I risultati confermano l’efficacia dell’approccio proposto e il suo potenziale per migliorare i sistemi di manutenzione basata sulle condizioni nel settore ferroviario.
Identification of wheel flat defects via axle box accelerations: experimental field test on a railway circuit
HOXHA, ENKELED
2023/2024
Abstract
This thesis contributes to the development of a diagnostic algorithm for detecting wheel defects—specifically, wheel-flat defects—in rail systems, paving the way for integration into predictive maintenance protocols. Wheel-flat defects are discrete irregularities in wheel geometry, typically resulting from sliding during braking or insufficient rail adhesion, which produce periodic vibrations and accelerate wear on both rolling stock and infrastructure. An advanced sensor node was designed to acquire axle-box vertical acceleration data, from which key features were extracted. The system was validated through a test campaign on the San Donato circuit in Bologna, Italy. Sensitivity analyses examined how defect dimensions, the number of defects, vehicle speed, track layout, and axle load affect diagnostic performance. A binary classification algorithm was then developed to distinguish between defective and non-defective wheel states. The results confirm the effectiveness of the proposed approach and its potential to enhance condition-based maintenance in the rail industry.File | Dimensione | Formato | |
---|---|---|---|
2025_04_Hoxha_Master_Thesis.pdf
non accessibile
Descrizione: Testo tesi
Dimensione
19.61 MB
Formato
Adobe PDF
|
19.61 MB | Adobe PDF | Visualizza/Apri |
2025_04_Hoxha_Executive_Summary.pdf
non accessibile
Descrizione: Testo executive summary
Dimensione
1.1 MB
Formato
Adobe PDF
|
1.1 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/10589/235099