Automatizirano upravljanje sustavima korištenjem senzorskih podataka

master's thesis

• Title (english): System Management Using Sensor Data
• Author: Nino Malbašić
• Mentor: Saša Dešić (mentor)
• Committee member: Saša Dešić (predsjednik povjerenstva)
• Committee member: Gordan Ježić (član povjerenstva)
• Committee member: Mario Kušek (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2017-07-12, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Internet stvari je popularni koncept koji omogućuje povezivanje senzorskih sustava ili uređaja i stvari koje se svakodnevno koriste, pružajući mogućnost da stvari komuniciraju međusobno te s višim slojevima sustava slanjem i primanjem podataka. Algoritmi strojnog učenja nude optimizirane metode analize i obrade velikih količina podataka te su vrlo pogodni za unošenje inteligencije u sustave. Kombinacija velike količine podataka koju produciraju senzorski sustavi te algoritama učenja unutar sustava interneta stvari mogu stvoriti vrlo kvalitetne sustave koji mijenjaju način života čovjeka ili industrije u kojoj su primijenjeni. Fokus ovog rada je na razvoju pametnog sustava koji će veliku količinu podataka prikupljenih sa senzorskih sustava iskoristiti u svrhu automatiziranog upravljanja sustavom. Kao konkretan primjer implementiran je sustav pametnog termostata koji uči čovjekove navike tijekom zadnjih 30 dana korištenja te je sposoban preuzeti automatsku kontrolu nad podešavanjem temperature na termostatu. Implementacija se sastoji od nekoliko dijelova: Mongo baze podataka, podatkovnog modela, upravljača pametnog termostata, algoritma učenja te korisničkog sučelja. Mongo baza podataka popunjena je senzorskim podacima koje za potrebe ovog rada simulira korisnik promjenom stanja termostata. Centralni upravljač pametnog termostata brine o komunikaciji s bazom podatka, pokretanju i obradi rezultata algoritma učenja te posluživanju korisničkog sučelja. Implementacija algoritma učenja u obliku je algoritma k – medoida.

Abstract (english)

Internet of things is a popular concept that provides connectivity between sensors or devices and everyday objects, enabling them to communicate with each other or with higher layers by sending and receiving data. Machine learning algorithms provide optimized data analytics and processing methods. They are very suitable for implementing artificial intelligence into systems. A combination of large sets of data, that are produced by sensors, and learning algorithms, implemented in the IoT systems, can create systems of great quality. Also, it can bring life-changing effects for humans or an industry where they are applied. The focus of this thesis is a development of a smart system which will use a large set of data collected from sensors and its purpose is managing automated systems. The concrete implementation is a smart thermostat system which learns human habits for last 30 days and it can take automatic control of thermostats temperature. The implementation consists of several parts: Mongo database, data model, smart thermostat controller, learning algorithm and user interface. Mongo database with sensors data which is, for this purpose, simulated with users’ actions of changing thermostat value. Central smart thermostat controller takes care of communication with the database, starting learning algorithm, result analysis and serving data for the user interface. Learning algorithm contains an implementation of k – medoids clustering.

• Keywords: Internet stvari (IoT); Umjetna inteligencija; Strojno učenje; Analiza i obrada velike količine podataka; Automatizirano upravljanje sustavima; Algoritam k – medoida; Pametni termostat
• Keywords (english): Internet of things (IoT); Artificial intelligence; Machine learning; Data analysis and processing; Automated system management; K – medoids algorithm; Smart thermostat
• Language: croatian
• URN:NBN: urn:nbn:hr:168:304635
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2019-03-13 22:30:51