Aline Neves Da Silva,Mahelvson Bazilio Chaves,Filipe Alves Coelho,Frede de Oliveira Carvalho, Universidade Federal de Alagoas, Centro de Tecnologia, aline.neves10@gmail.com
Abstract:
Changings in anthropical and natural forcings of the hydrological system at the Mundaú river has been the main reason by the recurrence of extremely hydrological events, which have harmed so deeply the population of Alagoas and Pernambuco. The understanding and knowledgement of the rainfall-runoff transformation of Mundaú basin river is a relevant tool that may subsidize actions wich reduce this prejudicial events to the riverine population. Thechniques based on Artificial Neural Network (ANR) have been used efficiently to estimate temporal series, due mainly the capacity to modeling non-linearity. Among the ANRs, stands the Multi-Layer Perception type, whose the three-layer structure allows to find any relationship between data, making this network able to perform such predictions. Then, using the MLP network, monitored flows of 1974-2006, was observed the applicability of the intelligent systems for large basins like Mundaú river.
Palavras-chave: rio Mundaú, redes neurais, previsão hidrológica.