Machine Learning on the Edge

Examensarbete på grundnivå

Please use this identifier to cite or link to this item: https://hdl.handle.net/20.500.12380/250158
Download file(s):
File Description SizeFormat 
250158.pdfFulltext13.77 MBAdobe PDFView/Open
Type: Examensarbete på grundnivå
Title: Machine Learning on the Edge
Authors: LÖVGREN, JOHAN
Olsson, Anton
Abstract: Machine learning is a tool for data analysis which can construct a model of a system without necessarily requiring deeper insight into the system. This model can then be used for analysis of the system. This type of analysis is of great interest for use with so called predictive maintenance; to be able to discover an abnormality in the behaviour in the system which can lead to the system breaking, before it actually happens. A platform for machine learning was provided by Ekkono and used to predict the state of a small DC motor. The prediction used data from various sensors to gain information about the current state of the system. This data was used to predict how the state would change a short time in the future. Fault detection was not within the scope of this project, we only concern ourselves with predicting how the values read from the sensors would change. In order to achieve this, a system which could be used for data collection was constructed. The system consisted of an Arduino Uno which collected data about the DC motor using various sensors. This data was analysed using a Raspberry Pi. The prediction was also done on the Raspberry. The accuracy of the predictions was then analysed.
Keywords: Informations- och kommunikationsteknik;Data- och informationsvetenskap;Information & Communication Technology;Computer and Information Science
Issue Date: 2017
Publisher: Chalmers tekniska högskola / Institutionen för data- och informationsteknik (Chalmers)
Chalmers University of Technology / Department of Computer Science and Engineering (Chalmers)
URI: https://hdl.handle.net/20.500.12380/250158
Collection:Examensarbeten på grundnivå // Basic Level Theses



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.