Machine learning methods for predictive maintenance using real-time data and time-frequency analysis - Diploma Thesis

Aug 3, 2026

This diploma thesis compares predictive maintenance techniques based on machine learning to estimate the Remaining Useful Life of a wind turbine shaft ball bearing. Classification algorithms, degradation models, and real-time updates are evaluated, including a Kalman Filter approach for continuous state refinement. The work first studies bearing failure mechanisms, the fundamentals of predictive maintenance, and machine learning methods. It then implements and assesses multiple prediction workflows, concluding with an efficiency comparison of each method.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages92

Reading Time3 h 52 min

AccessFree to download

Introduction

Problem Statement

Objectives

Bearing failure

Predictive Maintenance

What predictive maintenance is

Comparison to other maintenance approaches

PM applications in industry

Condition Monitoring

Machine Learning

Data Analysis and Feature Engineering

Classification algorithms

UNIVERSITY OF WEST ATTICA UNIVERSIDAD DEL PAIS VASCO FACULTY OF ENGINEERING FACULTY OF ENGINEERING VITORIA-GASTEIZ DEPARTMENT OF MECHANICAL ENGINEERING DEPARTMENT OF MECHANICAL ENGINEERING Diploma Thesis Machine learning methods for predictive maintenance using real-time data and time-frequency analysis Dimitrios Iason Papadopoulos Supervisors: Georgios Chamilothoris, Vanessa Garcia, Saioa Etxebarria Vitoria-Gasteiz, May 2023 GRADO EN INGENIERÍA MECÁNICA TRABAJO FIN DE GRADO Machine learning methods for predictive maintenance using real-time data and time-frequency analysis Alumno/Alumna: PAPADOPOULOS DIMITRIOS IASON Director/Directora (1): GARCIA VANESSA Director/Directora (2): ETXEBARRIA SAIOA Fecha: 29-05-2023 Acknowledgments First of all, I would like to thank my supervisor from the mechanical engineering department of the University of West Attica, Georgios Chamilothoris, for guiding me throughout the duration of my thesis and my supervisors from the mechanical engineering department of the University of the Basque Country, Vanessa Garcia and Saioa Etxebarria for their important contribution in my effort. Moreover, I would like to thank my family and friends, for supporting me for the whole duration of my studies. Abstract In this diploma thesis, different techniques of Predictive Maintenance based on Machine Learning are compared. In particular, the Remaining Useful Life of a ball bearing of the shaft of a Wind Turbine was predicted with different methods: Classification algorithms, degradation models and real time updates using a Kalman Filter. In the first half, the theory of ball bearing failure mechanisms, predictive maintenance and machine learning is analyzed. At the second half, different methods are implemented for the prediction of the remaining useful life. Last, the writer comes to a conclusion about the efficiency of each method. Key words: Machine Learning, Predictive Maintenance, Remaining Useful Life, Degradation Models, Classification, Kalman Filter Περίληψη Σε αυτή τη διπλωματική εργασία, γίνεται σύγκριση μεθόδων μηχανικής μάθησης για προδεικτικήσυντήρηση . Ειδικότερα, γίνεται πρόβλεψη για την εναπομένουσα ωφέλιμη ζωή ενός εδράνου,στον άξονα μιας ανεμογεννήτριας, με τις εξής μεθόδους, αλγόριθμους classification, μοντέλα degradation και συνεχής ανανέωση, με χρήση του φίλτρου Kalman. Σε πρώτη φάση, αναλύεταιη θεωρία, σχετικά με τους μηχανισμούς αστοχίας των εδράνων, την μηχανική μάθηση και τηνπροδεικτική συντήρηση . Στη συνέχεια, αυτές οι μέθοδοι, χρησιμοποιούνται για τον υπολογισμότης εναπομένουσας ωφέλιμης ζωής . Τέλος, ο συγγραφές αναλύει τα συμπεράσματά του για τηναποδοτικότητα της κάθε μεθόδου . Key words: Μηχανική μάθηση, Προδεικτική συντήρηση, Εναπομένουσα ωφέλιμη ζωή, Degradation Models, Classification, Φίλτρα Kalman Table of Contents Abstract ........................................................................................................................................................ 4 Περίληψη...................................................................................................................................................... 5 Table of figures ........................................................................................................................................... 8 1.Introduction.............................................................................................................................................. 9 1.1 Problem Statement .......................................................................................................................... 9 1.2 Objectives ......................................................................................................................................... 9 1.3 Bearing failure .................................................................................................................................. 9 2.Predictive Maintenance..

FAQs

Which predictive maintenance methods are compared in the thesis?

The thesis compares classification algorithms, degradation models, and real-time updates using a Kalman Filter for Remaining Useful Life prediction.

What component is selected for Remaining Useful Life prediction?

The Remaining Useful Life is predicted for a ball bearing on the shaft of a wind turbine.

How is the thesis structured from theory to implementation?

The first part analyzes bearing failure mechanisms, predictive maintenance, and machine learning, while the second part implements prediction methods and concludes by comparing their efficiency.

Aug 3, 2026

Upload by Valentina