Título: Analysis of migration to the Brazilian free energy market based on statistical methods and artificial neural networks
Autores: Fausto Marques Rodrigues Junior, Guilherme Baptista Bastos, Marcos Cesar da Rocha Seruffo, Fernando Augusto Ribeiro Costa, Karla Figueiredo and Harold Dias de Mello Junior
Resumo:
Since 1995, the Brazilian government has regulated a differentiated model for selling electricity with the objective of stimulating free competition. Widespread in recent years, with less restrictive adhesion requirements, the so-called free market allows the consumer to customize the contract, contemplating price, demand and supply period, established in a flexible way. Thus, the customer has the freedom to negotiate according to his needs, weighing costs and benefits. This paper describes the migration process from the Regulated Contracting Environment (RCE) to the Free Contracting Environment (FCE) and the savings achieved by a consumer of subgroup A4 and green tariff mode in the concession area of Light, which operates in the state of Rio de Janeiro. Statistical methods and artificial neural networks were employed in forecasting the consumption series, in the period from 2016 to 2019, to identify the best contracting conditions. Specifically, the long and short-term memory networks (LSTM) obtained the best performance with validation error of less than 1%. With simulated RCE data, we show the financial advantage consumers would have if they had not yet migrated to the FCE.
Palavras-chave:
Free energy market, time series forecasting, neural networks.
Páginas: 8
Código DOI: 10.21528/CBIC2021-109
Artigo em pdf: CBIC_2021_paper_109.pdf
Arquivo BibTeX: CBIC_2021_109.bib