Analysis of wind energy in Brazil using Time Series

• Autor(a) principal: Silva, Mateus dos Santos
• Data de Publicação: 2022
• Outros Autores: Santos, Pedro Henrique Alves Bittencourt, Santos, Ricardo Vitor Ribeiro dos, Nascimento, Mateus do, Pascoa, Marcelino Alves Rosa de, Pereira, Renato Nunes, Oliveira, Tiago Almeida de
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Research, Society and Development
• Texto Completo: https://rsdjournal.org/index.php/rsd/article/view/24827
• topic: Wind energy; Trend; Forecasts; SARIMA.; Energía eólica; Tendencia; Previsiones; SARIMA.; Energia eólica; Tendência; Previsões; SARIMA.

This paper aim was, adjust a time series model to the generated electrical energy series through the wind grid for studying trend, seasonality and making predictions. The used historic series consists in the electric energy generated through the wind grid, collected monthly by the Operador Nacional do Sistema Elétrico. The range is between jan/2007 until mar/2021, with 171 observations. The series was divides into two groups, first (jan/2007 to dec/2019) used for modelling process (calibration) and the other one (jan/2020 to mar/2021) for predictions evaluation (validation). For making predictions it was used apr/2021 to dec/2022 period. In the procedures, first was applied a Box-Cox transform on data scale for turn the model into additive. Then, the presence of trend was checked. From the transformed original series with an order 1 difference, FAC and FACP correlograms, were possible purpose some models. The residual not correlated and a lower AIC were the criteria used for the models. From the chose ones, were made prediction for jan/2020 to mar/2021 period, that were compared to the real observations through EQMP. The SARIMA (5,1,2)×(0,0,3)12 model was chosen because of its lowest EQMP. Other observation is related to the next months followed a rising pattern since 2015. The purposed model for forecasting the amount of electric energy generated by the wind grid will help the managers, giving them time for programming the proper energy’s distribution.