Inteligentna dijagnostika kvarova rotacijske opreme male brzine vrtnje

master's thesis

• Title (english): Intelligent fault diagnosis in low-speed rotating machinery
• Author: Domagoj Kežman
• Mentor: Dragutin Lisjak (mentor)
• Granter: University of Zagreb; Faculty of Mechanical Engineering and Naval Architecture; Zagreb
• Defense date and country: 2019-07-11, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Mechanical Engineering

Abstract

Održavanje tehničkih sustava u industriji jedna je od najvažnijih djelatnosti svakog poduzeća. Kako se tehnologija razvija, tako i procesi održavanja postaju sve važniji te se ciljevi održavanja redefiniraju od otklanjanja kvarova nakon nastanka prema poduzimanju mogućih radnji za izbjegavanje nastanka kvara koji bi prouzročio havariju. Za to se može koristiti strategija održavanja po stanju, opisana u ovom radu. Održavanje po stanju je simulirano mjerenjem i prikupljanjem podataka o vibracijama rotacijske opreme male brzine vrtnje na simulatoru kvarova. Cilj rada je bio na temelju prikupljenih podataka sa simulatora kvarova razviti računalni model koji će procjenjivati potencijalne kvarove na rotacijskoj opremi male brzine vrtnje. Za modeliranje je korištena konvolucijska neuronska mreža. Promatralo se kako promjene određenih hiperparametara utječu na točnost i vrijeme treniranja mreže. Ukupno su istrenirane 144 konvolucijske neuronske mreže te točnost u najboljoj mreži iznosi 100 %, a u najlošijoj 51 %.

Abstract (english)

Maintenance of technical systems in the industry is one of the most important activities in a company. As technology develops, maintenance processes become more important because the maintenance goals are being redefined from eliminating failures after their occurrence to taking actions to avoid the occurrence of a fault that can cause breakdowns. To do this companies use condition based maintenance strategy described in this paper. Condition based maintenance was simulated by measuring and collecting vibration data of small speed rotation equipment on the fault simulator. The aim of this thesis was based on collecting the data from the failure simulator in order to develop a computer model that will evaluate potential faults in low-speed rotating machinery. A convolutional neural network was used for modeling. It was observed how changes in certain hyperparameters affect the accuracy and time needed to train the network. A total of 144 convolutional neural networks were trained and the accuracy in the best network was 100 % and in the worst 51 %.

• Keywords: održavanje po stanju; duboko učenje; konvolucijske neuronske mreže; strojno učenje; vibracije
• Keywords (english): condition based maintenance; deep learning; convolutional neural networks; machine learning; vibrations
• Language: croatian
• URN:NBN: urn:nbn:hr:235:199959
• Study programme: Title: Mechanical Engineering; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka strojarstva (magistar/magistra inženjer/inženjerka strojarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2019-07-04 09:42:34