Uma Análise Empírica de Redes Neurais MLP Aplicada à Previsão de Vazões (An Empirical Analysis of MLP Neural Networks Applied to Streamflow Forecasting)

Danilo Braga de (danilobraga@bcc.unifal-mg.edu.br), Mariana Dehon Costa e (mariana@bcc.unifal-mg.edu.br), Ricardo Menezes (ricardo@bcc.unifal-mg.edu.br)


Federal University of Alfenas
This paper appears in: Revista IEEE América Latina

Publication Date: June 2011
Volume: 9,   Issue: 3 
ISSN: 1548-0992


Abstract:
Nowadays, in Brazil there is a large energy potential that comes from hydro mineral sources, which most part of the electricity consumed comes from this source. According to this, it is important emphasize that the decision-making related with planning of the operation of the reservoirs of hydroelectric plants has been done based mainly on preview knowledge of the flow. Thereby, this work aims to conduct an exploratory study about the Artificial Neural Networks type MLP to estimate which is the best setting to perform the stream flow forecast. This study was applied to the Grande River basin, in addition, with the achieved results, it was possible to observe that the search of appropriate parameters shows significant gains in the execution of the forecasts and can to reduce the error level obtained.

Index Terms:
Streamflow Forecasting, Power Systems Planning, MLP Neural Networks Configuration, Time Series Forecasting.   


Documents that cite this document
This function is not implemented yet.


[PDF Full-Text (687)]