Utilização das redes neurais artificiais para redução da subjetividade do método AHP aplicado a vulnerabilidade ambiental: estudo de caso do Córrego do Fundo, Carmo do Paranaíba - MG

• Autor(a) principal: Barbosa, Gustavo Rodrigues
• Data de Publicação: 2015
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UFU
• Texto Completo: https://repositorio.ufu.br/handle/123456789/14227; https://doi.org/10.14393/ufu.di.2015.367
• topic: Inteligência artificial; Bacias hidrográficas; Carmo do Paranaíba (MG) - Recursos naturais; Vulnerabilidade ambiental; Redes neurais artificiais; Bacia hidrográfica; Environmental vulnerability; Artificial neural networks; Watershed; CNPQ::ENGENHARIAS::ENGENHARIA CIVIL

Many researches have applied the Analytical Hierarchical Process (AHP) technique created by Saaty (1980), in the multicriterial analysis related to management of water and environmental resources. Among these works, highlight the vulnerability analysis proposed by Ross (1994) which became a reference for many other researchers in Brazil. The use of AHP technique has gained importance in environmental studies because of its association with geographic information systems to assign different degrees of susceptibility classes to morphogenesis and pedogenesis variables on a map. What is not appear in the literature is an assessment of the influence of experience and other human conditions in the initial consideration of the technical implementation and that may incur subjectivity in the analysis. Therefore, this research proposes to apply the method of artificial neural network (ANN) as a way to refine the weights indicated by experts in the application of AHP technique and minimize the subjectivity of the analysis process. In this research the AHP technique results are ponderations ass proposed by Ross (1994), by local experts through questionnaires and the application of ANN in determining the map of environmentally vulnerable areas of the watershed Córrego do Fundo, Carmo do Paranaíba, MG (Brasil). The results allow affirming applicability of RNA's to reduce the subjectivity of the data obtained with the AHP method has relevant consistency.