The present study was to develop a model on Artificial Neural Networks (ANN) in order to estimate the oxygen dissolved in the water of the river Alegria, located in Medianeira in the state of Paran. The model was developed based on data from the river water quality over the study interval. For training and validation of the model were generated 132 data groups: with 22 collections in 6 seasons. The input variables in the network were the water quality parameters except the (OD), which set as output. Given the results of the simulations carried out in order to predict the concentration of oxygen dissolved in the river water, depending on the number of variables involved, with an average error of 11, 42% can be concluded that a neural network can be used to predict the available oxygen in the waters of a river.
展开▼