Detection of Faults in Electrical Panels Using Deep Learning Method (CROSBI ID 653684)
Prilog sa skupa u zborniku | izvorni znanstveni rad | međunarodna recenzija
Podaci o odgovornosti
Mlakić, Dragan ; Nikolovski, Srete ; Baus, Zoran
engleski
Detection of Faults in Electrical Panels Using Deep Learning Method
In the image analysis, a big trend within the field of artificial intelligence is using the Deep Learning method, which is an upgrade of the existing neural network adaptive architecture (ANN). Deep Learning is a major new field in machine learning that encompasses a wide range of neural network architectures designed to perform various tasks. In the thermography energy sector, examples that are processed on a daily basis are sampling of active energy components, focus segmentation, and fault classification. The most popular network architecture for Deep Learning in image analysis is the convolution neural network (CNN), where traditional machine learning methods require determination and calculation, from which the algorithm training comes. Deep Learning approach captures important features as well as the appropriate weight of these attributes to make decision for new data. This paper describes a method and tool that are available to build and conduct an effective analysis of the Deep Learning Method for electrical components.
Deep Learning ; electrical components ; thermal imaging ; picture analysis ; Convolution Neural Network Introduction
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
Podaci o prilogu
55-61.
2017.
objavljeno
Podaci o matičnoj publikaciji
Proceedings of International Conference on Smart Systems and Technologies 2017 (SST 2017)
Drago Žagar, Goran Martinović, Snježana Rimac Drlje, Kruno Miličević
Osijek: Fakultet elektrotehnike, računarstva i informacijskih tehnologija Sveučilišta Josipa Jurja Strossmayera u Osijeku
978-1-5386-3776-0
Podaci o skupu
International Conference on Smart Systems and Technologies 2017 (SST 2017)
predavanje
17.10.2017-20.10.2017
Osijek, Hrvatska