guest ::
| Home > Conference materials > Papers > Testing a statistical forecasting model of electric energy consumption for two regions in the Czech Republic |
Original title:
Testing a statistical forecasting model of electric energy consumption for two regions in the Czech Republic
Authors: Rajdl, Kamil ; Farda, Aleš ; Štěpánek, Petr ; Zahradníček, Pavel
Document type: Papers
Conference/Event: Global Change: A Complex Challenge /4th/, Brno (CZ), 2015-03-23 / 2015-03-24
Year: 2015
Language: eng
Abstract: Precise forecasting of electric energy consumption is of great importance for the electric power industry. It helps system operators optimally schedule and control power systems, and even slight improvements in prediction accuracy might yield large savings or profits. For these reasons, many forecasting models based on various principles have been developed and studied. Because of energy consumption’s strong dependence on weather conditions, such models often utilize outputs from numerical weather prediction models. In this study, we present and analyse a statistical model for forecasting hourly electrical energy consumption by customers of E.ON Energie, a.s. in two regions of the Czech Republic. The aim of this model is to create hourly predictions up to several days in advance. The model uses hourly data of consumed energy from 2011–2014 and corresponding predictions of temperature and cloudiness provided by the ALADIN/ CZ model. The statistical model is based on a regression analysis applied to appropriate data samples and supplemented by several optional post-processing methods. Specifically, we use a robust linear regression algorithm to identify energy consumption’s dependence on temperature, the meteorological variable with the largest influence on consumption. Our post-processing methods focused on removing prediction bias resulting from economic situations (represented by the goss domestic product, GDP) and sudden temperature changes. We analysed the presented model from the point of view of the hourly predictions’ accuracy for 2013 and 2014. Accuracy was primarily measured by mean absolute error. It was evaluated for individual months, and the effects of individual parts of the model on accuracy value are shown. Introduction
Keywords: Czech Republic; electric energy; forecasting model
Project no.: LO1415 (CEP)
Funding provider: GA MŠk
Host item entry: Global Change: A Complex Challenge : Conference Proceedings, ISBN 978-80-87902-10-3
Institution: Global Change Research Institute AS ČR (web)
Document availability information: Fulltext is available at the institute of the Academy of Sciences.
Original record: http://hdl.handle.net/11104/0256911
Permalink: http://www.nusl.cz/ntk/nusl-202717
The record appears in these collections:
Research > Institutes ASCR > Global Change Research Institute
Conference materials > Papers
Authors: Rajdl, Kamil ; Farda, Aleš ; Štěpánek, Petr ; Zahradníček, Pavel
Document type: Papers
Conference/Event: Global Change: A Complex Challenge /4th/, Brno (CZ), 2015-03-23 / 2015-03-24
Year: 2015
Language: eng
Abstract: Precise forecasting of electric energy consumption is of great importance for the electric power industry. It helps system operators optimally schedule and control power systems, and even slight improvements in prediction accuracy might yield large savings or profits. For these reasons, many forecasting models based on various principles have been developed and studied. Because of energy consumption’s strong dependence on weather conditions, such models often utilize outputs from numerical weather prediction models. In this study, we present and analyse a statistical model for forecasting hourly electrical energy consumption by customers of E.ON Energie, a.s. in two regions of the Czech Republic. The aim of this model is to create hourly predictions up to several days in advance. The model uses hourly data of consumed energy from 2011–2014 and corresponding predictions of temperature and cloudiness provided by the ALADIN/ CZ model. The statistical model is based on a regression analysis applied to appropriate data samples and supplemented by several optional post-processing methods. Specifically, we use a robust linear regression algorithm to identify energy consumption’s dependence on temperature, the meteorological variable with the largest influence on consumption. Our post-processing methods focused on removing prediction bias resulting from economic situations (represented by the goss domestic product, GDP) and sudden temperature changes. We analysed the presented model from the point of view of the hourly predictions’ accuracy for 2013 and 2014. Accuracy was primarily measured by mean absolute error. It was evaluated for individual months, and the effects of individual parts of the model on accuracy value are shown. Introduction
Keywords: Czech Republic; electric energy; forecasting model
Project no.: LO1415 (CEP)
Funding provider: GA MŠk
Host item entry: Global Change: A Complex Challenge : Conference Proceedings, ISBN 978-80-87902-10-3
Institution: Global Change Research Institute AS ČR (web)
Document availability information: Fulltext is available at the institute of the Academy of Sciences.
Original record: http://hdl.handle.net/11104/0256911
Permalink: http://www.nusl.cz/ntk/nusl-202717
The record appears in these collections:
Research > Institutes ASCR > Global Change Research Institute
Conference materials > Papers
Record created 2016-02-22, last modified 2022-09-29