APPLYING ARTIFICIAL INTELLIGENCE TO PAVEMENT DESIGNS

• Autor(a) principal: Conceição Farias do Egito Costa, Stephanny
• Data de Publicação: 2024
• Outros Autores: Elísio de Figueiredo Lopes Lucena, Adriano, de Paiva, William
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Revista Baru - Revista Brasileira de Assuntos Regionais e Urbanos
• Texto Completo: https://seer.pucgoias.edu.br/index.php/baru/article/view/13338
• topic: MeDiNa; Resilient Modulus; Artificial Neural Networks; MeDiNa; Módulo resiliente; Redes Neurales Artificiales; MeDiNa; Módulo de resiliência; Redes Neurais Artificiais

In Brazil, for at least six decades, pavement designs have been regulated by an empirical method based on the North American experience, which sometimes did not meet the Brazilian reality, triggering premature wear of the highways. This wear promotes increased fuel consumption, vehicle deterioration, a decrease in transportation speed and user safety. Given this scenario, the new Método de Dimensionamento Nacional, MeDiNa, of a mechanistic-empirical nature, developed by Brazilian engineers and researchers, arose. The MeDiNa requires parameters referring to the mechanical characteristics of materials, including the resilient behavior of soils. For this, the repeated load triaxial test must be performed, but the necessary equipment requires large investment capital and specialized labor. Given the possibility of acquiring reliable results through intelligent systems, in this work artificial neural networks were developed to predict the resilient modulus of soils in the state of Goiás, based on their physical characterization. The prediction of the resilient modulus obtained good results indicating the viability of using this tool and consequent facilitation of the use of MeDiNa in the design of pavements.