Título: Previsão de Vazões Naturais Médias Mensais Usando uma Rede Neural Nebulosa Adaptativa
Autores: Ballini, Rosangela; Soares, Secundino; Andrade, Marinho Gomes
Resumo: Analysis and forecast of seasonal streamflow series are of utmost importance in the operation planning of water resources systems. One of the greatest difficulties in forecasting of those series is the seasonality nature of streamflow series due to dry periods of the year. This work suggests the application of neurofuzzy network model to seasonal streamflow forecasting. This model learns membership functions parameters for each input variable from training data, processes data automatically adjusted to cover the whole input space. The problem of seasonal streamflow forecasting is solved using a database of average monthly inflows of one Brazilian hydroelectric plant located at Grande River. Comparison of the neurofuzzy model with multilayer neural network and periodic autoregressive models are also included to illustrate the performance of the approach. The results show that the models here proposed provide a better performance than the other ones considering one-step-ahead forecasting and multi-step-ahead forecasting, with forecasting errors significantly lower than the other approaches.
Código DOI: 10.21528/CBRN2001-088
Artigo em pdf: 5cbrn_088.pdf
Arquivo BibTex: 5cbrn_088.bib