Potencial strojnega učenja za identifikacijo napak v dinamiki rotorjev

Balkovec, Gregor (author), Slavič, Janko (mentor)

Obravnavan je potencial strojnega učenja pri identifikaciji napak v ležajih. V začetku je predstavljeno stanje znanosti na tem področju, katere metode strojnega učenja se uporabljajo in njihova uspešnost. Nato so predstavljene osnove najpogostejših metod. Sledi implementacija teh metod na novem naboru podatkov, ki predstavlja bolj realne pogoje vgradnje ležajev. Na koncu so metode primerjane med seboj.

• Language: Slovenian
• Keywords: strojno učenje, umetna inteligenca, k-bližnjih sosedov, metoda podpornih vektorjev, večslojni perceptron, python, scikit-learn
• Work type: Bachelor thesis/paper
• Typology: 2.11 - Undergraduate Thesis
• Organization: FS - Faculty of Mechanical Engineering
• Place of publishing: Ljubljana
• Publisher: [G. Balkovec]
• Year: 2020
• Number of pages: XXII, 44 str.
• PID: 20.500.12556/RUL-120212
• UDC: 004.85:621.822(043.2)
• COBISS.SI-ID: 30104067
• Publication date in RUL: 17.09.2020

Secondary language

• Language: English
• Title: The potential of machine learning for fault identification in rotor dynamics

Abstract

The potential of machine learning for fault identification in bearings is discussed. Firstly, the current state in the field is presented, which machine learning methods are most commonly used and their applicability. Secondly, the basics of selected methods are presented. Finally, the implementation of selected methods on a bearing dataset is discussed. In the end, the methods are compared with each other.

• Keywords: machine learning, artificial intelligence, k-nearest neighbor, support-vector machine, multilayer perceptron, python, scikit-learn