Loading...
Publication
Previsão multi-atributo do preço no Mercado Ibérico de Eletricidade
| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 1.63 MB | Adobe PDF | |||
| 1.2 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
A eletricidade tem vindo a adquirir uma maior presença nas nossas vidas e estima-se que o futuro seja cada vez mais elétrico.
Hoje em dia temos acesso a quantidades enormes de dados que acabam por não ter grande valor acrescentado se não puderem ser utilizados para suportar tomadas de decisão ou
planear antecipada e corretamente sistemas.
As previsões são instrumentos fundamentais para apoiar tomadas de decisão.
A eletricidade é considerada uma commodity muito especial, pois embora tenha havido
diversos progressos relativos ao desempenho de baterias, a eletricidade é, largamente, um
bem não armazenável. Nesse sentido, o preço de eletricidade apresenta características únicas
que torna a sua previsão uma tarefa difícil.
Acreditamos ser possível recorrer a dados disponíveis na Internet para efetuar previsões de
preços de eletricidade que possam ser usadas por decisores no setor.
Neste trabalho apresentamos um processo de previsão quantitativa e, por forma a
compreender as previsões de preço de eletricidade, investigamos a previsão multi-passos e
multi-atributo.
Consideramos diversas séries temporais de dados disponíveis abertamente em diversas
fontes. Os dados utilizados inserem-se em cinco categorias: cronológicos, preço, procura,
produção e clima.
O estudo compreende três intervalos de tempo 1 de janeiro de 2019 (00:00) a 31 de dezembro
de 2019 (23:00), 1 de janeiro de 2018 (00:00) a 31 de dezembro de 2019 (23:00) e 1 de
janeiro de 2010 (00:00) a 31 de dezembro de 2019 (23:00).
Aos resultados, aplicamos um método de apoio à tomada de decisão multi-atributo, o
TOPSIS.
Electricity has been acquiring a greater presence in our lives and it is estimated that the future will be increasingly electric. Nowadays we have access to enormous amounts of data that end up not having much added value if they cannot be used to support decision making or plan systems in advance and correctly. Forecasts are key tools to support decision-making. Electricity is considered a very special commodity, because although there have been several advances regarding the performance of batteries, electricity is largely a non-storable commodity. In this sense, the price of electricity presents unique characteristics that make its forecasting a difficult task. We believe it is possible to resort to data available on the Internet to make electricity price forecasts that can be used by decision makers in the sector. In this paper we present a quantitative forecasting process and, in order to understand electricity price forecasting, we investigate multi-step and multi-attribute forecasting. We consider several time series of openly available data from various sources. The data used fall into five categories: chronological, price, demand, production and weather. The study comprises three time intervals 1 January 2019 (00:00) to 31 December 2019 (23:00), 1 January 2018 (00:00) to 31 December 2019 (23:00) and 1 January 2010 (00:00) to 31 December 2019 (23:00). To the results, we applied a multi-attribute decision support method, TOPSIS.
Electricity has been acquiring a greater presence in our lives and it is estimated that the future will be increasingly electric. Nowadays we have access to enormous amounts of data that end up not having much added value if they cannot be used to support decision making or plan systems in advance and correctly. Forecasts are key tools to support decision-making. Electricity is considered a very special commodity, because although there have been several advances regarding the performance of batteries, electricity is largely a non-storable commodity. In this sense, the price of electricity presents unique characteristics that make its forecasting a difficult task. We believe it is possible to resort to data available on the Internet to make electricity price forecasts that can be used by decision makers in the sector. In this paper we present a quantitative forecasting process and, in order to understand electricity price forecasting, we investigate multi-step and multi-attribute forecasting. We consider several time series of openly available data from various sources. The data used fall into five categories: chronological, price, demand, production and weather. The study comprises three time intervals 1 January 2019 (00:00) to 31 December 2019 (23:00), 1 January 2018 (00:00) to 31 December 2019 (23:00) and 1 January 2010 (00:00) to 31 December 2019 (23:00). To the results, we applied a multi-attribute decision support method, TOPSIS.
Description
Keywords
Mercado Ibérico de Eletricidade (MIBEL) Eletricidade Previsão Preços Automated Machine Learning Decisão Multi-Atributo Price forecasting Iberian Electricity Market (MIBEL) AutoML Multi-Attribute Decision
Citation
Peres, Gonçalo Martins - Previsão multi-atributo do preço no Mercado Ibérico de Eletricidade [Em linha]. [S.l.]: [s.n.], 2021. 2 vols.