Previsão de demanda de energia elétrica com redes neurais artificiais e análise por série de Fourier

• Ano de defesa: 2016
• Autor(a) principal: Evandro Lopes de Oliveira
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Modelo de previsão; Redes neurais; Demanda de energia; Série de Fourier; Engenharia elétrica; Redes neurais (Computação); Fourier, Séries de
• Link de acesso: http://hdl.handle.net/1843/BUBD-AN6PSF

The current model of the electricity market in Brazil is due to the breaking up of the state monopoly on the electrical energy sector. This model has regulations that divide the energy sector in sections that offer the service, creating a competitive environment. The trend in the market that competes for electricity is that companies seek ways to provide the customers need. In order to improve the quality of service provided,companies analyze market behavior, check the critical points and elaborate control and intervention strategies in these points.Considering that the electricity demand planning in Brazil is strategic, this work shows some demand forecasting methodologies using socio-economic performance indicators, weather indicators and electricity demand history. The proposed models are based on artificial neural networks, statistical algorithms, trend identification and analysisof exogenous variables. The application of the forecasting methods, demonstrated in this work, enables the achievement of a future behavior of the National Interconnected Systems electrical demand with an average close to its historical serie.