Kratkoročno predviđanje potrošnje električne energije

master's thesis

• Title (english): Short term load forecasting
• Author: Marko Deak
• Mentor: Domagoj Jakobović (mentor)
• Committee member: Domagoj Jakobović (predsjednik povjerenstva)
• Committee member: Marko Čupić (član povjerenstva)
• Committee member: Leonardo Jelenković (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2015-07-16, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Problem kratkoročnog predviđanja potrošnje električne energije jedan je od osnovnih primjera prognoze vrijednosti vremenskih sljedova te ima značajnu ulogu u električnoj opskrbi. Svrha ovog rada je demonstrirati primjenu metoda genetskog programiranja na problemu kratkoročnog predviđanja potrošnje električne energije te usporedba rezultata sa postojećim dobivenim primjenom drugih metoda. Uz osnovnu implementaciju genetskog programiranja dodano je i linearno skaliranje te semantičko genetsko programiranje za poboljšanje dobivenog rješenja te strože poštivanje semantike prisutne u ulaznim podatcima. Najveći dio optimizacije dobivenih modela implementacija je metoda skupnog učenja primijenjenih na genetsko programiranje, odnosno algoritama GPBoost i BCC. Prikazani su rezultati implementiranog algoritma te njegovih različitih varijacija uz utjecaj promjene parametara algoritma na kvalitetu najboljeg dobivenog rješenja. Konačno su dana razmatranja rezultata te kraće ideje za daljnja poboljšanja rezultata.

Abstract (english)

The short-term electric energy load forecasting problem is one of the basic examples of time series forecasting and has an important role in electric power distribution. The goal of this paper is to demonstrate the application of genetic programming methods on the problem of short-term load forecasting and to compare results with the existing ones gained when using other methods. Together with the basic implementation of genetic programming linear scaling and semantic genetic programming was added to improve the current solution and enforce the stricter following of the semantics which exist in the input data. The largest part of optimization of current models was the implementation of ensemble learning methods applied to genetic programming, namely algorithms GPBoost and BCC. Results of the implemented algorithm and its different variations are presented together with the effect that changing of algorithm parameters has on the quality of the best solution. Finally, further consideration of results is given as well as short ideas for further improvement of the results.

• Keywords: genetsko programiranje; kratkoročno predviđanje potrošnje; semantičko GP; skupno učenje; linearno skaliranje
• Keywords (english): genetic programming; short-term load forecasting; semantic GP; ensemble learning; linear scaling
• Language: croatian
• URN:NBN: urn:nbn:hr:168:784624
• 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: 2018-02-06 08:24:52