Felipe de Jesús Ruiz López

Transporte de Contaminantes en Aguas Subterráneas mediante Redes Neuronales Artificiales

An Artificial Neural Network model was developed for predicting pollutants transport (cupper and cadmium) in saturated, homogeneous and isotropic media for several textural classes. The models were trained and evaluated from the equation proposed by Ogata and Banks that considers advective and diffusive terms. Backpropagation structures were developed, using a three-layer architecture considering 4, 7 and 10 neurons in the hidden layer. For training and simulation the LevenbergMarquardt algorithm was used, the Log-sigmoid transfer function was applied in the hidden layer and a linear function was applied in the output layer. The results demonstrate that the Artificial Neural Network is a useful mathematical tool; it has low computational requirements and allows estimating the transport of pollutants in saturated, homogeneous and isotropic media.