Izrada automatiziranog sustava kalkulacija korištenjem neuronskih mreža

master's thesis

• Author: Petar Branislav Jelušić
• Mentor: Ante Poljičak (mentor)
• Committee member: Lidija Mandić (član povjerenstva)
• Committee member: Ante Poljičak (član povjerenstva)
• Committee member: Maja Strgar Kurečić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Graphic Arts; Zagreb
• Defense date and country: 2016-7-5, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Graphic Technology

Abstract

Umjetne neuronske mreže su računalni modeli zasnovani na modelu bioloških neuronskih mreža. Imitacijom tokova informacija i signalnih puteva u ljudskom mozgu, umjetne neuronske mreže pokušavaju izračunati i predvidjeti razne matematičke probleme i funkcije. Posljednjih nekoliko desetljeća u konstantom su razvoju, a njihova primjena sve je šira. U teoretskom dijelu rada objasnit će se princip funkcioniranja umjetnih neuronskih mreža. Nakon toga, u eksperimentalnom dijelu kronološki će se objasniti proces izrade automatiziranog sustava tiskarskih kalkulacija – od prikupljanja podataka iz radnih naloga, do implementacije tih podataka u računalni program MATLAB. Upravo će se u MATLAB-u koristiti neuronske mreže. Krajnji cilj izrade automatiziranog sustava kalkulacija jest preciznije izračunavanje potrebne količine materijala pri svakom tiskarskom procesu, kako bi se izbjegli problemi (višak proizvoda zauzima mjesto u skladištu tiskare) i troškovi (manjak proizvoda stvara dodatni trošak, budući da se proces tiska mora započeti još jednom) u tiskarskoj proizvodnji. Navedeni proces izrade kalkulacija bit će potkrijepljen pripadajućim fotografijama i prikazima zaslona.

Abstract (english)

Artificial neural networks are computational models based on biological neural networks. Using neural networks' structure and functions, artificial neural networks can compute and predict various mathematical tasks. In the past decades, artificial neural networks are improving constatntly, and their application is growing rapidly. At the beginning of the research, the general functions and applications of artificial neural networks will be explained briefly. In the experimental part, making of an automated printing calculation will be chronologically demonstrated – from gathering the information, to importing the data into MATLAB software. MATLAB will be used to work with neural networks. The goal of having an automated printing calculation is predicting the exact amount of paper material needed for every printing process. That way, it's possible to avoid various problems (excess of material takes up precious storage space) and expenses (insufficient amounts of material creates more expenses, as the printing process has to start over for the second time). Each process within the research will be explained with appropriate photos and screenshots.

• Keywords: biološke neuronske mreže; umjetne neuronske mreže; perceptron; MATLAB; Neural Network Toolbox; tiskarske kalkulacije
• Keywords (english): biological neural networks; artificial neural networks; perceptron; MATLAB; Neural Network Toolbox; printing calculations
• Language: croatian
• URN:NBN: urn:nbn:hr:216:550284
• Study programme: Title: Graphic technology; specializations in: Graphic technology, Design of graphic products; Course: Graphic technology; Study programme type: university; Study level: graduate; Academic / professional title: sveučilišni/a magistar/magistra inženjer/inženjerka grafičke tehnologije (sveučilišni/a magistar/magistra inženjer/inženjerka grafičke tehnologije)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2024-04-18 21:09:30