Rede neural artificial aplicada à previsão de vazões médias mensais da bacia hidrográfica do rio Piancó.

Abstract

The streamflow forecasting in a water system is one of the techniques used to minimize the impact of the uncertainties of the climate on the administration of the water resources. That technique can be considered one of the principal challenges related to the integrated knowledge of the climatology and of the hydrology of the river basin. The aim of this work was it of modeling the no-lineal relationship between rainfall and streamflow in the Pianco river basin, in the paraibano semiarid, using the technique of Artificial Neural Networks (ANN). Here the capacity of ANN was evaluated to model the process rainfallrunoff in monthly base. It was considered, during the training of ANN, the network architecture and, weights initialization influence. In the end of the training it was chosen the best architecture, to model the streamflow monthly mean in the studied basin, with base in the acting of the model. The architecture of ANN that produced better result was RC315L with values for the determination coefficient, efficiency coefficient and standard estimate error (SEE) equal to 92.0%, 77.0% and 8.29 respectively.