Avtomatska zaustavitev naprave za stepanje smetane s pomočjo metod strojnega učenja

WEISSENBACH, JAN (author), Sadikov, Aleksander (mentor)

Diplomska naloga rešuje problem avtomatskega zaustavljanja naprave za stepanje smetane. Problem rešuje z uporabo algoritmov strojnega učenja na podlagi podatkov obremenjenosti motorja. Smetane imajo med seboj zelo različne karakteristike, zato jih je težko prepoznati in napovedati zaustavitev procesa. Opisane so že obstoječe rešitve za omenjen problem, med drugimi uporaba nevronskih mrež in uporaba klasičnega algoritma. Opisani so podatki in ugotovitve, na podlagi katerih smo se odločili za gradnjo petih napovednih modelov glede na maso. Opisane so značilke in razlog zakaj smo jih uporabili pri učenju samih modelov. Prav tako so predstavljeni rezultati omenjenih modelov.

• Language: Slovenian
• Keywords: Strojno učenje, klasifikacija, MATLAB, časovna serija.
• Work type: Bachelor thesis/paper
• Typology: 2.11 - Undergraduate Thesis
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2021
• PID: 20.500.12556/RUL-129675
• COBISS.SI-ID: 76574211
• Publication date in RUL: 07.09.2021

Secondary language

• Language: English
• Title: Automated stopping of the cream-whipping machine using machine learning

Abstract

This thesis solves the problem for automatic stop of a device for whipping cream. The problem is solved using classic machine learning algorithms based on motor load. Creams have different characteristics and are hard to recognise and it is even harder to predict the stopping point. It describes already existing solutions such as usage of neural networks and classic algorithm. We have described used data and findings based on which we have created five different classification models. We have described used statistic metrics and presented model results.

• Keywords: Machine learning, classification, MATLAB, time-series.