Open Journal Systems

APLICABILIDADE DE REDES NEURAIS ARTIFICIAIS PARA ANÁLISE DE GERAÇÃO DE ENERGIA DE UM SISTEMA FOTOVOLTAICO CONECTADO A REDE ELÉTRICA

Elisangela Pinheiro, Adalberto Lovato, Ricardo Ruther

Resumo


Fontes de energia flutuantes, como a solar, estão aumentando sua participação na matriz energética. Concomitantemente está aumentando a dependência dessa energia e consequentemente, novos métodos de previsão de suprimento precisam ser desenvolvidos. Redes neurais artificiais têm sido usadas para prever a irradiação solar com sucesso. No entanto seu uso na previsão da energia futura disponível não tem sido explorado. O algoritmo de Levenberg-Marquard foi usado em diversas configurações que foram treinadas com a aplicação da função de transferência logssigmoide na camada oculta e purelin na camada de saída. Os dados de entrada foram a irradiação solar, a temperatura ambiente, a temperatura dos módulos solares e a hora do dia. Os dados de target foram os valores da energia produzida. O treinamento foi realizado com 5, 10, 15, 20, 25, 30 e 60 neurônios na camada oculta. O melhor resultado foi obtido com 30 neurônios, com coeficiente de correlação de 0,98.

 

Palavras-chave: Energia solar. Redes neurais artificiais. Simulação da produção de energia. Algoritmo de Levenberg-Marquard.


Texto completo:

PDF

Referências


BOUSFIELD, P. M. Uso de uma Rede Neural para a Previsão do Volume de Lodo Gerado em estações de tratamento de efluente: Um estudo de caso, 2008. Universidade da região de Joinville. Disponível em:.

CHATZIAGORAKIS.P., GEORGOULAS. PAPADOPOULOU, S. ELMASIDES,C., GIAOURIS, D., Seferlis, P., Sirakoulis, CH.G., KARAFYLLIDIS, I., PAPADOPOULOS, A. I., STERGIOPOULOS, F., ZIOGOU, C., IPSAKIS, D., Voutetakis, S. e ANDREADIS, I. Application of Neural Networks Solar Radiation Prediction for Hybrid Renewable Energy Systems. Engineering Applications of Neural Networks, v. 459, p. 133-144, 2014.

CHOW, S. K. H.; LEE, E. W. M.; LI, D. H. W. Short-term prediction of photovoltaic energy generation by intelligent approach. Energy and Buildings, v. 55, p. 660–667, 2012. Elsevier B.V. Disponível em: .

FIORIN, D. V; MARTINS, F. R.; SCHUCH, N. J. Aplicacões de redes neurais e previsões de disponibilidade de recursos energéticos solares. Revista Brasileira de Ensino de Física, v. 33, n. 1, p. 1309 1 – 1309 20, 2011.

GILMAN, A. Dobos; KASBERG, M. P50/P90 Analysis for Solar Energy Systems Using the System Advisor Model, NREL, May 13-17, 2012. Disponível em:< http://www.nrel.gov/docs/fy12osti/54488.pdf>.

HAGAN, Martin T.; DEMUTH, Howard B.; BEALE, Mark Hudson; JESÚS, Orlando De. Neural Network Design. 2ª Ed. eBook, 2015.

MEKKI, A., ALOUI, F., DHOUIB, A., SAYADI, S. Effects of Phanerochaete chrysosporium on biologic activity of soil amended with olive mill wastewaters. Journal of Soil Science and Environmental Management, v. 3, n.1, p. 1-8, 2012.

MELLIT, A.; KALOGIROU, S. A.; HONTORIA, L.; SHAARI, S. Artificial intelligence techniques for sizing photovoltaic systems: A review. Renewable and Sustainable Energy Reviews, v. 13, n. 2, p. 406–419, 2009. Disponível em: .

MELLIT, A.; KALOGIROU, S. A. ANFIS-based modelling for photovoltaic power supply system: A case study. Renewable Energy, v. 36, n. 1, p. 250–258, 2011. Elsevier Ltd. Disponível em:.

MELLIT, A.; KALOGIROU, S. A. Artificial intelligence techniques for photovoltaic applications: A review. Progress in Energy and Combustion Science, v. 34, n. 5, p. 574–632, 2008.

NREL. National Renewable Energy Laboratory: Research Cell Efficiency Records. 2015. Disponível em: . Acesso em: junho de 2016. http://www.nrel.gov/ncpv/images/efficiency_chart.jpg

RÜTHER, R. e DACOREGIO, M., 2000. Performance Assessment of a 2 kWp Grid-connected, Building-integrated, Amorphous Silicon Photovoltaic Installation in Brazil. Prog. Photovolt. Res. Appl. (8), pp. 257 - 266.

RÜTHER, R., 1998. Experiences and operational results of the first grid-connected, building-integrated, thin-film photovoltaic installation in Brazil. WCPEC2 - 2nd World Conference on Photovoltaic Solar Energy Conversion, Vienna, Austria.

YADAV, A. K.; CHANDEL, S. . Artificial Neural Network based Prediction of Solar Radiation for Indian Stations. International Journal of Computer Applications, v. 50, n. 9, p. 1–4, 2012.




DOI: http://dx.doi.org/10.5380/rber.v6i5.48431

Apontamentos