Predikcija emisija štetnih plinova vozila korištenjem algoritama strojnog učenja s podatcima prikupljenih OBD2 uređajem

master's thesis

• Title (english): Prediction of vehicle emissions using machine learning algorithms with data collected by the OBD2 device
• Author: Tin Vaiti
• Mentor: Tonči Carić (mentor)
• Committee member: Edouard Ivanjko (predsjednik povjerenstva)
• Committee member: Pero Škorput (član povjerenstva)
• Committee member: Miroslav Vujić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Transport and Traffic Sciences; (Division of Intelligent Transport Systems and Logistics); (Department of Intelligent Transport Systems); Zagreb
• Defense date and country: 2022-09-22, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Traffic and Transport Technology; Intelligent Transport Systems and Logistics

Abstract

Konstantnim povećanjem broja vozila na cestama javlja se sve veći problem emisija štetnih ispušnih plinova. Navedeno je izazvalo globalni interes za primijenjeno istraživanje u područjima analize podataka i strojnog učenja nad podacima prikupljenim s vozila. Kako bi se olakšalo praćenje količine emisija štetnih ispušnih plinova razvijeni su različiti modeli procjene štetnih emisija. U ovom diplomskom radu izraden je model predikcije emisija ugljikovog dioksida korištenjem metoda strojnog učenja. Za treiranje modela bili su korišteni javno dostupni podaci koji se sastoje od tehničkih podataka o samom vozilu, procijenjene emisije ugljičnog dioksida za nova laka vozila te ocjene potrošnje goriva specifične za odredeno vozilo. Korišteni alat za izradu modela bio je programski jezik Python u kojemu se obradila predobrada podataka, vršilo se testiranje pet različitih regresija od kojih je regresija s povećanjem gradijenata ispala najpouzdanija i koja se koristila za testiranje modela na privatno prikupljenim podacima. Za testiranje modela koristili su se osobno prikupljeni podaci pomoću OBD-II (eng. On-Board Diagnostics, OBD) uređaja. Rezultati primjene modela bili su prikazani na manjem dijelu rute gdje je model uspješno predvidio količinu emisija u svakom dijelu rute koristeći ulazne parametre prikupljene OBD-II uredajem.

Abstract (english)

With the constant increase in the number of vehicles on the roads, the problem of harmful gas emissions is emerging. This has sparked global interest in applied research in the areas of data analysis and machine learning on data collected from vehicles. In order to facilitate the monitoring of the amount of harmful gas emissions, various models for the assessment of harmful emissions have been developed. In this thesis, a carbon dioxide emission prediction model was developed using machine learning methods. Publicly available data consisting of technical data about the vehicle itself, estimated carbon dioxide emissions for new light vehicles, and vehicle-specific fuel consumption ratings were used to train the model. The tool used to create the model was the Python programming language in which the data was processed, five different regressions were tested, of which the regression with increasing gradients turned out to be the most reliable and which was used to test the model on privately collected data. To test the model, personally collected data using an OBD-II (eng. On-Board Diagnostics, OBD) device was used. The results of the model application were shown on a smaller part of the route where the model successfully predicted the amount of emissions in each part of the route using the input parameters collected by the OBD-II device.

• Keywords: predviđanje štetnih emisija vozila; strojno učenje; obradivanje podataka; OBD-II
• Keywords (english): prediction of harmful vehicle emissions; machine learning; data processing; OBD-II
• Language: croatian
• URN:NBN: urn:nbn:hr:119:906095
• Study programme: Title: Intelligent Transport Systems and Logistics; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka prometa (magistar/magistra inženjer/inženjerka prometa)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2022-12-09 17:02:07