Institutional Repository of Vilnius University
Lietuvos elektros energijos paklausos prognozavimas /
Abstract
dc:descriptionThe goal of this research is to create electricity demand models using time series methods, compare them and pinpoint the model with the lowest forecasting error. In the course of the research, trends in electricity consumption, seasonality and external factors affecting fluctuations in electricity demand are identified. The studied time series are characterized by: annual seasonality - electricity consumption increases in the winter months, in the summer demand reaches its lowest values, weekly seasonality - electricity is consumed more on weekdays than on weekends, daily seasonality - consumption reaches its lowest values at night, in the 7 - 10 period demand reaches daily peak. Based on the literature examined, interactions between time and weather indicators (temperature, relative air humidity, cloud cover and wind speed) are taken into account during the analysis. 4 models were tested: SARIMA, dynamic harmonic regression, SARIMAX and dynamic harmonic regression with exogenous variables. When building the models optimal trainig is identified based on the errors of the predicted values. The lowest prediction errors are obtained using model SARIMAX - ARIMA(7, 1, 0)(2, 1, 2)[24]X with the training sample starting at 2022.03.24. Exogenous variables of the model are temperature, wind speed, cloudiness, relative air humidity, dummy, time variables are hour, day of the week, month, holiday and interactions between the listed variables.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2023
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Šerpatauskaitė, Monika,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD192842384&prefLang=en_US