| | SLO | ENG | Cookies and privacy

Bigger font | Smaller font

Show document Help

Title:Analiza učinkovitosti različnih metod strojnega učenja pri napovedovanju sončnega sevanja : diplomsko delo
Authors:ID Popović, Dragan (Author)
ID Igrec, Dalibor (Mentor) More about this mentor... New window
Files:.pdf UN_Popovic_Dragan_2024.pdf (6,31 MB)
MD5: 10BA150F9DCE05250AEDEC5638D3EA6D
 
Language:Slovenian
Work type:Bachelor thesis/paper
Typology:2.11 - Undergraduate Thesis
Organization:FE - Faculty of Energy Technology
Abstract:V diplomskem delu smo primerjali učinkovitost dveh modelov strojnega učenja, LSTM (Long Short-Term Memory) in FFNN (Feedforward Neural Network), pri napovedovanju globalnega horizontalnega obsevanja (GHO). Sončna energija je ključna za trajnostni razvoj, saj predstavlja čisto in obnovljivo alternativo fosilnim gorivom, vendar njeno izkoriščanje zahteva natančne napovedi sončnega obsevanja. V nalogi smo najprej obravnavali teoretične osnove sončne energije, fotovoltaike in izzive, povezane z napovedovanjem vrednosti GHO. Nato smo v praktičnem delu izvedli simulacije v okolju MATLAB, kjer smo modela trenirali na podatkih iz let 2018–2021 in napovedovali vrednosti za leto 2022. Rezultati so pokazali, da je LSTM model splošno boljši pri napovedovanju GHO, saj je izkazal večjo stabilnost, natančnost in zmožnost zajemanja sezonskih sprememb v primerjavi s FFNN modelom, ki je imel več težav pri napovedovanju ekstremnih vrednosti in sezonskih nihanj. Na podlagi teh ugotovitev zaključujemo, da je LSTM model primernejši za napovedovanje GHO, medtem ko FFNN model potrebuje nadaljnjo optimizacijo. Naloga ponuja prispevek k razvoju natančnejših napovednih modelov, ki so ključni za optimizacijo izkoriščanja sončne energije in prispevek k trajnostnemu razvoju.
Keywords:sončna energija, strojno učenje, globalno horizontalno obsevanje
Place of publishing:Maribor
Place of performance:Krško
Publisher:[D. Popović]
Year of publishing:2024
Number of pages:XVIII, 78 f.
PID:20.500.12556/DKUM-90463 New window
UDC:[662.997+621.383.51]:004.85(043.2)
COBISS.SI-ID:225522179 New window
Publication date in DKUM:07.02.2025
Views:189
Downloads:24
Metadata:XML DC-XML DC-RDF
Categories:FE
:
Copy citation
  
Average score:(0 votes)
Your score:Voting is allowed only for logged in users.
Share:Bookmark and Share



Hover the mouse pointer over a document title to show the abstract or click on the title to get all document metadata.

Licences

License:CC BY-NC-ND 4.0, Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International
Link:http://creativecommons.org/licenses/by-nc-nd/4.0/
Description:The most restrictive Creative Commons license. This only allows people to download and share the work for no commercial gain and for no other purposes.
Licensing start date:04.09.2024

Secondary language

Language:English
Title:Analysis of the effectiveness of different machine learning methods for predicting solar radiation
Abstract:This thesis compares the effectiveness of two machine learning models, LSTM (Long Short-Term Memory) and FFNN (Feedforward Neural Network), in predicting global horizontal irradiance (GHI). Solar energy is crucial for sustainable development as it provides a clean and renewable alternative to fossil fuels, but its utilization requires accurate predictions of solar irradiance. The thesis first explores the theoretical foundations of solar energy, photovoltaics, and the challenges associated with predicting GHI values. In the practical part, simulations were conducted in MATLAB, where the models were trained on data from 2018–2021 and used to predict values for 2022. The results showed that the LSTM model is generally superior in predicting GHI, demonstrating greater stability, accuracy, and capability in capturing seasonal variations compared to the FFNN model, which struggled with predicting extreme values and seasonal fluctuations. Based on these findings, we conclude that the LSTM model is more suitable for GHI prediction, while the FFNN model requires further optimization. This thesis contributes to the development of more accurate predictive models, which are essential for optimizing the use of solar energy and supporting sustainable development.
Keywords:solar energy, machine learning, global horizontal irradiance


Comments

Leave comment

You must log in to leave a comment.

Comments (0)
0 - 0 / 0
 
There are no comments!

Back
Logos of partners University of Maribor University of Ljubljana University of Primorska University of Nova Gorica