Multi-label classification of a hydraulic system using Machine Learning methods.

No Thumbnail Available
Journal Title
Journal ISSN
Volume Title
Perustieteiden korkeakoulu | Master's thesis
Date
2022-07-29
Department
Major/Subject
Exchange studies in Computer Science
Mcode
SCI3042
Degree programme
Master’s Programme in Computer, Communication and Information Sciences
Language
en
Pages
51
Series
Abstract
In this project, a condition monitoring of a hydraulic system has been developed. The research consisted of a health categorization of the most relevant physical and non-physical elements of the system. The objective has been to use different ML models to classify the state of the elements in each cycle and to be able to know through the information of the features of each cycle whether an element of the system needs to be replaced or not and also find out the work efficiency of each element of study. This research therefore follows a supervised learning analysis in which two types of classifications will be carried out. The first one will be a multiclass classification done with different ML techniques that will try to classify the categories of each label separately, getting to know for each cycle which is the state of the analyzed element. On the other hand, a multilabel analysis will follow. In this case, all labels will be taken, and different performances will be done. The main objective in this chapter will be to elaborate different tests with different ML models in order to see which is the optimal one for this system, and which is the algorithm that should be used to monitor this type of hydraulic system. In addition to these classification analyses, the correlation between the different data will be assessed beforehand in order to verify relationships or coincidences that may be relevant.
Description
Supervisor
Jung, Alexander
Thesis advisor
Jung, Alexander
Keywords
multilabel classification, monitoring, hydraulic system, supervised learning, multiclass classification
Other note
Citation